Mastering Advanced ChatGPT Prompt Engineering

1. Why I Fell in Love with AI & Prompt Engineering final: Yan hide. Welcome back. Today, I want to talk about how I went from Office WTekkit to having deep into the world of AI and prompt engineering. If you have ever felt that excitement when you first discovered technology or dreamed of future full of Pots and AI, you will probably relate to my story. So let's rewind a bit. Ever since I was a kid, I have been curious about how technology works, and I remember watching all these science fiction movies and thinking, Wow, imagine if I could live in the future, surrounded by AI and advanced tech. I dreamed of having personal AI assistant like Javas, someone who could handle everything for me. In school, I was really into building things. I even tried to make my own robot around sixth and seventh grade. I took the motor from the toy car, and I attached it to DIY robot and thought, This is it. This is going to fly. But instead of it just started walking backward. And I love. After experimenting with batteries and wiring, I finally got it moving forward. Though sometimes in circle, I kept adding features like a magnet for picking up metal particles. I added that in the lego robot and a sponge I'm to pick up dust, and soon enough, I had this DIY cleaning robot. Even entered into school science fair project and got third place. I got third prize because they said it looked more like a DI project than a real world robot, but everyone was talking about my projects. Actually, I don't have that DII project DIY robot I made in school time that used to clean the room and all that. I made it in 2008 or ten when I was in six standard or seven standard. And now, actually, but now I have one robot. I want to show you, but I also don't want to show you because the boxes I used, they are not inppriate like they are not supposed to show you on camera till I will give you the glimpse of how it looks. Okay, so this is the motor. I attached it to robot, and this is the robot. I place the motor over here and added battery in backwards, and it was moving forward as well as it was rotating. And if I change the battery, like if I supply current oppositely, it used to move back. So I used to do experiment. I added magtside it, as well as sponge and all that. Um, so everyone was, like, shocked, like, how I created this when I was in school. I also have a video let me show the video. I want to show you that video, but I can't show you directly. So I will show you like this. I hope you understand why I'm not able to show you this. Okay, I hope you are able to see. As you can see, it is rotating. And if I change the batteries, it will move back into power. So this is how it used to work. Growing up, I didn't have personal computer, so I spent hours in cyber capes, fixing computer issue for my friends. It was my thing. I was the guy everyone calls when something went wrong with mobile or computer. And from that moment, I knew I wanted to build something even bigger, a humanoid air robot. But as I learned in college, building robots can get super expensive, so I decided to dive into programming instead. That's when I discovered Python in 2016, and I realized Hey, I could actually use this to make AI La Javers. So I started coding nonstop, making games, latter programs, a liter software. By the third year of my engineering, I was fully into motion learning. And by my final year, I was working on huge AI project that could predict things and even recommend products. That's when I really started seeing how powerful AI could be. Then one day I heard about this cool new tool. HatiPety I had to try and let me tell you, it blew my mind. I couldn't believe what it could do. It could write, it could code. It could even act like anyone I wanted. I was so hooked with this platform with this tool that I started experimenting with day and night, trying every trick and every prompt. Eventually, I found out there was an actual field dedicated to getting the best responses out of AI. That's prompt engineering. Realized, Wait, this is exactly what I have been doing all along. So I had go even deeper learning everything I could find on this new skill and developing my own techniques. Now, here I am sharing what I have learned with all of you. I feel like that dream I had as a kid living in the future has actually come true. AI is our future, and it's here right now. And the fact that I get to help people unlock the power and build their own amazing products is incredible. Honestly, seeing what my students have created, how they are using prompt engineering in their own lives and career makes everything worth it. I'm so excited to be part of this AI journey with you, and I hope this course help you create amazing things, too. Some of my student by learning prompt engineering and Chi Gibt from me, they was able to create chat boards. There is a one person called Div and she submitted one project. I was watching that project, and she mailed a chat board, a small chat board in ha Jebeti. It used to tell her a skincare routine, like what kind of skincare routine she should be having or let's say, if you add a picture of someone, if that person have some skin issue, then that chatbard used to tell you should apply this kind of serum as well as moisturize sunscreen. So I was so amazed, like, how students are learning prompt engineering and applying those skills to make real world projects. As well as there was one person who built a completely project, a website and made it live, I guess. And he also shared that with me. And the website was looking so amazing. And I guess it was three, four pages website, and he did this everything just writing prompts. And I know for that you have to have the knowledge of coding. You have to have knowledge about SMLCSS JoScript. Then you can upload those website, or you can make them live. There are so many projects they have done. I'm so proud of my student. They are able to do this kind of project and sharing those projects with me. I feel so lucky that I'm able to do something. I'm able to share my knowledge with them and they are able to create this kind of project. Honestly, seeing what my students have created, how they are using prompt engineering in their own lives and career makes everything worth it. I'm so excited to be part of this journey with you, and I hope this course help you creating amazing things too. So if you are ready to dive into prompt engineering and see what AI can really do, let's get started. Trust me, it is going to mean amazing journey. So that's what I. And this is my whole journey up until now. I also did lots of thing, but this was a small part of my life, and I really wanted to share this with you. I hope you got it but from West. I know there is nothing to get inspiration, but I just want to share, how I got into AI and all that. So that's what, and I will see when in the next is up. 2. Course Guide Navigating Examples, Assignments & Resources: Hi, everyone, Jane. Today, today, I want to talk about something special. So I hope you download that resources file. I want to talk about that. In this we'll talk about examples and sec assignments as well as projects. So we are going to talk all of them. So before we dive deeper, I want to take a moment to walk you through all the helpful resources, including this course. Following along with the course materials will make your learning experience smoother and more engaging and most importantly, more effective. So let's get into it. Inside the Z file you downloaded, you will find three key files and few extra to help keep you motivated along the way. The first file you will see is the example Pi. This contains all the examples I have shown you in the videos. You can use this example to practice the skills and techniques I have been demonstrating. In fact, there are even more examples in this pero. So if you want to extra practice, they're right here waiting for you. So name of that file is advanced prompt engineering tool kit and mastering every step. So this is the example file. And so, as you can see, so these are the examples I showed in this video while teaching, I was performing these kind of prompts, as well as if you scroll down a little bit, I added more examples. Um, you can try to experiment with them, and I encourage you to make your own prompt after doing this, after learning from this, experiment your own front, try to add more data into those front. So in this example, I level up the prompts. Like, first one is easy, then medium hard, very hard. So it is levelled like that. So, eventually, you will get comfortable. I solve easy, then if you are solving medium, then it will be easy for you. I don't want to give that like directly hard prompt. I want to um I I want to make sure you should go step by step when learning or when practicing it. Let's say if you are having trouble with any of this example, try watching the video again. Go step by step, then tackle the example on your own. And remember, if you're still having trouble, don't hesitate to reach out. I'm here to help. Now, next up, we have the assignments and project file. After each video, you will find an assignment or real world projects designed specially to reinforce what you have just learned. This file is divided into four levels of difficulty, easy, medium, hard, and very hard. So this is the project and assignment file. So in this I it assignments, as you can say, as well as the project, but I named them assignment, but these are project, real world projects. Um, all the instruction are project detail like this. So complete them, and after completing it, add project section. I already showed you how to do that. So after completing each project assignment, take a screenshot and upload it into project section. Don't copy paste link because links doesn't work. And some people provide different links, so I don't want that. So just copy paste, sorry, take a screenshot and paste it over there. These assignments are an amazing way to push your prompt engineering skill to new heights. Just take them one at a time, progressing at your own pace and submit each completed assignment in design section. Now onto the MCU file. Each topic in this course has its own set of multiple choice questions. These are perfect for testing your understanding as you go. After watching each video, come back to this file and give this question a try. They are designed to range from easy to hard, so you can track your progress as you move through this course. So this is the MCQ file, so as you can see, I add an MC here as well as the answer. Let's watch the video. Then come back to this file, read the MCQ and write to solid and also the answer is over here. So you can check your answer you are right or wrong. And I have also included a few extra resources to help you stay organized and motivated. So in this Z file, you will find some printable pages to help you track your daily learning tasks and progresses. You can stick them upon your wall or keep them on your desk, and there's a space for notetaking too. You can joon down thoughts as you watch, either digitally or on sheets. When you will extract that this file, you will see the kind of files inside it, we talked about SmentECQ and examples. They are over here. I also added gifts, and after that, I added this point sheet. So what this promotor technique is, like, let's say you plan, you want to study this prompting for 2 hours. Then after learning each topic or after learning for 25 minutes, take a break, a five minute break, and then again, love for 25 minutes, then take a break. And while doing that, you can mark this like event complete these circles. And I encourage you to print this out and keep on your desk, maybe stick on all and around it, because I always do that. If you see the wall in front sticky notes, and I just cross them like what kind of videos I want to make? What kind of social media post, I am to post or reels and all that. So I just stick it on the wall. It helps a lot. And after crossing it, it feels so good. Like, I just completed something. So I will highly encourage you to print this out and stick on your wall or you also press it on this. And I also added the note taking sheet over here. So, let's say you learn something like a personal pattern. And you can, like, Uh, print this out in topic, write the name of topic, and after that, write what you have learned? Oh, here, summarize it. And if you need one more page, I'll add it one more page. So print this out and write your notes. If you want to, if you have any specific notebook, then write in notebook as well. But I have observed that in my engineering when I used to study, if I learn any topic and I write it down, I used to grasp that knowledge a lot. I used to grasp that knowledge deeper, and I used to explain those to my student. And I also used to give my notes to my fellow friends. So that's why it is that's why it is effective. So write down in Notebook or if you want to print result, you can print this as well. And, if you're a digital person, if you want to add in notion or in note taking up, you can also do that. Finally, I have added a small gift, some exclusive wallpaper to keep you inspired and motivated. Feel free to share this wallpaper with your friends or set them as your background as a reminder of the journey you are going on. So if you open this gift and I added lots of wallpaper to motivate you. So as you can see, this one, prompt the future, you can set them as your wallpaper. Apto wallpaper or mobile wallpaper, it will remind you you have to learn prompt engineering, as well as it will keep you motivated and you will keep on the track. That's why I added. And in a few days I'm working on a book of prompt engineering. And when I will complete that book, I will also add it in this file. So in that book, I will add all the things I know about prompt engineering. I will also update this course when I learn new trick or if new update comes in, so I will update that as well. And just stay off of that book. When it well done, I will make an announcement. I will email you or something like that, and can go through that book and learn more about prompt engineering or new LLM models, as well as about new tricks and tapes, gel and all that. So stay off of that. And that book will be completely free for you, and it will be paired for other people. But you are my student, so that's why I'm giving that book totally free for you. So that's the quick overview of how to follow along with these course. Remember, this journey is you to take at your own piece. So dive into these resources whenever you need. I'm so excited to see the skills you have developed and the incredible projects you will create. So let's get started. So that's photo. I hope you downloaded G file, and if you're not, then download please download it. I will help you a lot. So that's what judo and I will see you guys in the next so. Oh 3. Skillshare Course Guide Navigating Examples, Assignments & Resources: Hi, everyone, he Daner. Today, today, I want to talk about something special. So I hope you download that resources file. I want to talk about that. In this we'll talk about examples and sec assignments as well as projects. So we are going to talk all of them. So before we dive deeper, I want to take a moment to walk you through all the helpful resources, including this course. Playing along with these course materials will make your learning experience smoother and more engaging and most importantly, more effective. So let's get into it. Inside the Z file you downloaded, you will find three key files and few extra to help keep you motivated along the way. The first file you will see is the example PD. This contains all the examples I have shown you in the videos. You can use this example to practice the skills and techniques I have been demonstrating. In fact, there are even more examples in this period. So if you want to extra practice, they're right here wedding for you. So name of that file is advanced prompt engineering tool kit and mastering every step. So this is the example file. And so, as you can see, so these are the examples I showed in this video while teaching, I was performing these kind of prompts, as well as if you scroll down a little bit, I added more examples. Um, you can try to experiment with them, and I encourage you to make your own prompt after doing this, after learning from this, experiment your own front, try to add more data into those front. So in this example, I level up the prawns. Like, first one is easy, then med them hard, very hard. So it is levelled like that. So um, like, eventually, you will get comfortable. I will solve easy, then, if you're solving medium, then it will be easy for you. I don't want to give that like directly hard prawns. I want to um I I want to make sure you should go step by step, when learning or when practicing it. Let's say if you are having trouble with any of this example, try watching the video again. Go step by step, then tackle the example on your own. And remember, if you're still having trouble, don't hesitate to reach out. I'm here to help. Now, next up, we have the assignments and project file. After each video, you will find an assignment or real world projects designed specially to reinforce what you have just learned. This file is divided into four levels of difficulty. Easy, medium, hard, and very hard. So this is the project and assignment file. So in this I edit assignments, as you can see, as well as the project, but I named them assignment, but these are project, real world projects. Um all the instruction are project detail like this. So complete them, and after completing it, add project section. I already showed you how to do that. So after completing each project assignment, take a screenshot and upload it into project section. Don't copy past link because links doesn't work. And some people provide different links, so I don't want that. So just copy paste, sorry, take a screenshot and paste it over there. These assignments are an amazing way to push your prompt engineering skill to new heights. Just take them one at a time, progressing at your own pace and submit each completed assignment in design section. Now onto the MCU file. Each topic in this course has its own set of multiple choice questions. These are perfect for testing your understanding as you go. After watching each video, come back to this file and give this question a try. They are designed to range from easy to hard, so you can track your progress as you move through this course. So this is the MCQ file, so as you can see, I add an MC here as well as the answer. Let's watch the video. Then come back to this file, read the MCQ and write the solid and also the answer is over here. So you can check your answer you are right or wrong. And I have also included a few extra resources to help you stay organized and motivated. So in this Z file, you will find some printable pages to help you track your daily learning tasks and progresses. You can stick them upon your wall or keep them on your desk, and there's a space for note taking too. You can joon down thoughts as you watch, either digitally or on sheets. When you will extract that this file, you will see the kind of files inside it, we talked about Assignment, CQ and examples. They are over here. A added gifts, and after that, I added this Pomodoro sheet. So what this Pomodoro technique is like, let's say you plan, you want to study this prompting for 2 hours. Then after learning each topic after learning for 25 minutes, take a break, a five minute break. And then again, love for 25 minutes, then take a while doing that, you can mark this like event, complete these circles. And I encourage you to print this out and keep on your desk, maybe stick on all and around it, because I always do that. If you see the wall in front of me, there are sticky notes, and I just cross them like what kind of videos I want to make? What kind of social media post, I am to post or reels and all that. So I just stick it on the wall. It helps a lot. And after crossing it, it feels so good. Like I just completed something. So I will highly encourage you to print this out and stick on your wall or also press it on desk. And I also added the note taking sheet over here. So, let's say you learn something like a personal pattern. And you can, like, print this out in topic, write the name of topic, and after that, write what you have learned over here, summarize it. And if you need one more page, I also add it one more page. So print this out and write your notes. If you want to, if you have any specific notebook, then write in noteboo as well. I have observed that in my engineering when I used to study, if I learn topic and I write it down, I used to grasp that knowledge a lot. I used to grasp that knowledge deeper, and I used to explain those to my student. And I also used to give my notes to my fellow friends. So that's why that's why it is effective. So write down in Notebook or if you want to print this, you can print this as well. And if you're a digital person, if you want to add in notion or in note taking up, you can also do that. Finally, I have added a small gift, some exclusive wallpaper to keep you inspired and motivated. Feel free to share this wallpaper with your friends or set them as your background as a reminder of the journey you are going on. If you open this gift and I added lots of wallpaper to motivate you, as you can see, this one, prompt the future, you can set them as your wallpaper. Apto wallpaper or mobile wallpaper, it will remind you you have to learn prompt engineering as well as it will keep you motivated and you will keep on the track. That's why I added. In few days I'm working on a book of prompt engineering. And when I will complete that book, I will also add it in this file. So in that book, I will add all the things I know about prompt engineering. I will also update this course when I learn new trick or if new update comes in, so I will update that as well. And just stay off of that book. When it well done, I will make an announcement. I will email you or something like that, and can go through that book and learn more about prompt engineering or new LLM models, as well as about new tricks and tips, agile and all that. So stay off of that. And that book will be completely free for you, and it will be paired for other people. But you are my student, so that's why I'm giving that book totally free for you. So that's the quick overview of how to follow along with these course. Remember, this journey is you to take at your own piece. So dive into these resources whenever you need. I'm so excited to see the skills you have developed and the incredible projects you will create. So let's get started. So that's a photo Zoo. I hope you downloaded this file, and if you're not, then download download it. I will help you a lot. So that's what Zoo, and I will see you guys in the next. Okay. 4. Why Making AI Write Like You Is Tricky in ChatGPT: Hi, everyone Chener. Today, we're diving to topic that might change how you think about AI and writing. Well, like, have you wondered, like, getting hagiBty write like you? It is way tricker than it seems. Well, buckle up because we are about to dive into fascinating world of a want prompt engineering. You know, once I try to get hadiBety to write a birthday card for my grandma, let's just say it ended up sounding more like former business letter than a warm personal note. That experience opened my eyes to challenge of AI writing, and I can't wait to share what I have learned with you. So in this video, we will explore why it is so hard to get AI to capture your unique voice. And we will look at some mind blowing ways people are tackling this challenge. Trust me, by the end of this video, you will never look at Ia generated text the same way again. So let's first understand the illusion of simplicity. At first glance, getting AI to write like might seem as easy as a pie, like just type in what you want and Walla. But here's the kicker. It's not that simple. Imagine you are trying to teach your pet parrot to talk like. You can just read a dictionary and accept it to sound exactly like you, right? The same goes for AI. It needs more than just words. It needs context, style and personality. For example, if you ask had Geppet to write a recipe for chocolate chip cookie, it might give you a perfect good recipe, but would it include your secret ingredient or that funny story about how you once accidentally use salt instead of sugar? Probably not. Now second point is the complexity of human writing. Now, let's dive a little deeper. Writing isn't just about stringling words together. It's an art form that reflects your unique experience, emotions, and thought process. Think about your favorite author. What makes their writing special? Maybe it's their witty humor, their vivid description or their ability to tug at your harsh strings. These qualities come from years of life experience, culture influence, and personal quirks. For instance, if you asked AI to write like J Rollins, it might rhyme and use silly words, but capturing the true essence of J Rollins, her imaginative words and subtle life lesson that a whole different ballgame. From JKR Rollins, I remember, I was, you know, I told you, like in previous course in Jack Jep prom Engineering course, I was writing a book. And that time I was doing some experiment and that time I thought, like, I will give my story. And it is about space and all that. And still I said to ha Jeopardy. Like, can you write this story, like how JK Rollins writes it? But ha Jeopardy messed up really bad. Like when I was reading that output, it kept saying, now I'm JC Rollins, and I'm writing like JJK Rollins. I was so confused, like, I don't want this. I want how JK Rollins writes it. Now that time I realize that we have to give context, style, personality, tone. So everything, then it will understand and then it will write like J Rollins. Now, third point is the power of prompt engineering. Here's where things get really exciting. This is the secret source that can help bridge the gap between generic air writing and your unique voice. Prompt engineering is like being a master shape. You are not just throwing ingredients into a pot. You are carefully crafting a recipe that guides the AI to create something special. It involves giving the AI specific instruction examples and context to the work with. Like, for example, instead of just saying write a story about a dog. So this is like generic prompt. You can you might say, write a heartwelm story about a loyal golden retire named Max. Help this elder owner overcome his loneliness. So use descriptive language and include dialogue that shows they are born, and you can see the difference. Mastering prompt engineering is like learning a new language. It takes time, practice, and a lot of trial and error. But when done right, it can produce some truly amazing results. So there you have it. Getting AI to write like you is hard because writing is complex, personal, and deeply human. But with the power of prompt engineering, we are getting closer to bridging that gap. Remember, AI is a tool, not a replacement for your unique voice. Is it to announce your creativity, not to replace it. Who knows with some practice in prompt engineering you might even discover new aspect of your writing style you never know existed. If you are integrated and want to learn more, check out upcoming videos on prompt Engineering or experience with different prompts in age pity. And, hey, why not share your experience in the command below? I love to hear about your adventure in AI writing. So in resources, I added examples which I use in this video, as well as some other examples you can try it on your own. And I will say, try to add your, like, own, let's say you want to write message to someone, and you make lots of typos or lots of grammar mistakes. So what you can do is that, um, let's say, you can teach to hat JPT like, write like this person and pass that message to hat JPET and see the output. Compare your input, your writing style and the output. I also added assignment like there are four, and there goes like this easy, medium, hard, and very hard. So try them and try to solve them. I also added the ulcer there in PLP, as well as in this platform. So go there, check that out, and I hope you understand what we are trying to do here. So that's a 42 days video. So stay curious, keep exploring and until next time, have a prompting. 5. How to submit your projects in Skillshare: Hi, everyone, hadnir. Today, I'm going to show you how you can submit your project in Scratch share. So let's let's say you watch this video. Which one? Okay, let's take this one. After watching the video, I hope you downloaded this file called Prompt Engineering plus Playbook. Download that file, and after opening that file, you will find some PDFs. So let me show you a few PDFs. So when you extract that resource file, you download it from Skillshare, you will get few PDFs as well as few images. So first images assignment. After that, we have MCQ. The answer is also included over here, and after that, I included some examples to practice or also few examples I actually tried on and lots of prawns in it. So go through this, all the examples, try it on your own, and complete that. And after that, I promost. So you can print this out. Let me show you. Okay, so you can print this promo dory technique. And let's say, let's say you want to learn prompt engining for 2 hours. So each day, you can divide 2 hours in 45 minutes. So you can, uh, learn for 25 minutes. After that, take five minute break, learn 25 minutes, mark this, and then again, take five break, then mark this. Then you have to continue that loop. And in this file, you get no taking sheets. So you can print it out and you can name you can add the name of coping. After that, like, you can add what you learned and add the write the summary. And you can also add it over here. Okay, let's start with assignments. So when you open this assignment, this is the front page, and we was talking about gameplay pattern. Okay, so we were talking about gameplay patterns. So Oh here, I added three assignments first easy, then middle and then hard. So just via all the tasks like what you have to do in this assignment or homework or in project. And we also add a description after that instruction and also add example. So you will get some inspiration as well as you will not get lost while doing this assignment. So after doing that in chat GPT, take a screenshot of that. And you can upload image over here, add project title. So in our case, across game play pattern and add some description like what you have done in this project, add image, video, or link. But mostly I recommend you to add an image from this. So add your screenshot over here. And if you want to make your project private, you can do that, if you don't want to show your project to your students or fellow classmates. So you can click this one. It will be only visible for you or not for me, but you have to share that project with me, so that's why don't click on this. And after submitting that screenshot or project or homework, just click on the publish. And after that, I can see your projects. I can create you or I can give you feedback. So that's how you can submit your project in skill share. So please download that resource file. It all the instruction like how you can submit your project over here and watch all the video, complete all the projects. I guess there are more than 40 or 50 projects or assignment, also homeworks. So do that. My approach in this class is learning by doing. I just want to give you more practical knowledge. By the end of this score, you will be skilled with prompt engining techniques. So that's all about how to submit your project in the cell. So that's main sign of thanks watching, and I will see you guys in the next 6. Crafting Effective Prompts and Clear Instructions in ChatGPT: Hey, amassing viewers. Do you felt you are speaking different language when you're talking with Cha Ji Petty? Well, Buccal, because today we are going to unlock the secret of getting what exactly you want from this air wonder. And, trust me, by the end of this video, you'll be chatting with Cha Jibeti like your old friend. So here's the funny story. Last week, I asked Hajipi to help me come up with a catchy slogan for my local community garden. I thought I was being clear, but instead of a friendly green them tag line, it gave me something that sounded like it was from iboty Company. Like cultivating tomorrow's innovation today. Not exactly what you had expected from neighborhood waggiPatg, right? But don't worry. I have learned from my mistake, and I'm here to share all the tips and tricks I have picked up along the way. So we are diving into the world of advanced prompts, instruction and writing in chat Jeopardy. You might ask me, hathan why does this matter? Because mastering this skill can turn hat Jeopardy from sometimes confusing chat body into your personal assistant, crave to your partner, or even your study pal. So stick around because we are about to explore three game changing techniques that will blow your mind and transform how you use chat Jeopardy. So first technique is the power of player instruction. So let's start with the basic giving clear instruction. Imagine you are asking your little sister to make you a sandwich. If you just say make a sandwich, who knows what you will end up with? So here's the real life hat jibe example. Let's say you want to plan a fun day out. Instead of asking, what should I do today? Try this. You can try prompt like this, like suggest five fun budget friendly activities for Sunny Saturday in a small town. You can also add your town name, O HM and suit for family with two kids aged eight to ten. If you're a family man, you can try this prompt if you are a bachelor like me, and you can modify this prompt. For now, let's go with this prompt and let's see what kind of output we get from Chat JP. So, as you can see, in Output, we got Pi because we asked for Pi force nature, scavage hunt, bike ride and picnic, D IFI mini golf, Visit Local Farm Market and Street Fair, or Dom Unit. Okay, so as you can see, these are like, really cool bland. If you want to go with family, then this is, like, really amazing. I could choose the third one. Sorry, the fourth one, visit the Local Farm Market and Street Fair. And fifth one is also goon. After that, we can go for the movie. Okay, so this is amazing. So you can see the difference. Like, if we ask, what should I do today? And if you ask this prompt, like suggest me five budget friendly activities on D Sunny Saturday, you can see the difference. If you ask the generate prompt, you will get generic answer. But if you add clear instruction, I know I didn't add the much instruction it. I just added few instruction till it gave me the really amazing output. The second prompt gives the hag pretty clear guidelines, helping it understand exactly what you're looking for. Now, next technique is the magic of context. Let's talk about context. This is like giving Chat GPID a backstory for your request. It helps the AI understand the bigger picture. For instance, if you are writing a story, don't just ask for ideas, set a scene. The generic prompt might look like this. Give me ideas for story. Instead, you can try. I'm writing a children bedtime story about ability total, and the story should teach the valuable perseverance. Can you suggest three possible adventure scenarios for my total zero? So as you can see, we added some context in it, like we want story about total, and we also want value like perseverance from this story. So that's the power of contexts. If I run this, I know there will be a huge story, and if you want to read it, you can pause the video, and then read it out. So we got, I guess, three stories away, the three little stories. Pause the video, read the story, and also try the previous prompt, like, give me the ILS for this story. But you have to try that prawn before this pmt. Like before giving this prompt, you have to try that pram because if you now give the generic prompt, still hat GPD will produce the amazing O put because from previous prompt, it will learn, we have to add context. So before giving that prompt, remember, you don't have to give any context like this one. Or hat GPD have memory, as you can see, memory updated, so it works on updated memories. So it learns on memory. So keep in mind, you have to try that prom prompt first. Give me ads for my story. After that, try this prom and then compare it out. So by providing context, you are guiding hat GPT to give you more relevant and thriller response. Now, here's where it gets really exciting. Sometimes Cha JBT's first response isn't quite right. That's okay. Think about it as a conversation. You can always ask for change or improvements. Let's say you asked AGP to explain how car engine works. But the explanation was too technical. Let's say you asked AGP to, like, explain how car engine works. So this is great output, and this is very technical. So I'm the person who likes car and all that. So for me, it is very easy to understand. But if there is one person who don't know anything about cars, then this output will be confusing to that person. So in this case, you can ask like this. That's great. But can you explain in a simple term as if you are talking to 10-year-old and use everyday objects as analogist to make it easier to understand. Okay, so we got the output. So let me read the points, cylinder and pistons, and it compared the objects with balloons and plunges the daily objects like we wrote in our com, use everyday objects as analogy. So hard protein, then it's work very well. After that, the four big step, the baking cookies Okay, that's nice. And if I read this after the explosion, all over left or gas need to go somewhere, okay? So as you can see, this really easy understanding explanation. If you want to read this output, pause the du, read this, then after that, read this one as well and compare on your own. Or I will add all the prompts in the resources, so you can try it on your own if you don't want to read this one. So this back and forth or iterative refinement is like sculping. You start with rough shape and keep refining until you get exactly what you want. And there you have folks, you have unlocked the secret to mastering chat Jeopardy prompts. So let's recap what we have learned. First one is give clear and specific instruction. Second one is provide context to set the scene. And third one is, don't be afraid to refine and ask for changes. Remember, communicating with ha Gibt is an art. It might take a little practice, but I promise it's worth it. Whether you are a student working on project, a professional seeking to create your ideas or just someone curious about AI, these tips will regionize how you use Jahipty. So if you want to learn more, try this technique out yourself and see the difference. And, hey, why not share your best Jaipty conversation in the comment below? I love to see what you're putting these tips into action. So as always, I added assignments, or examples, extra examples, try them, learn from them, so experiment with these examples, try to add your own story in it, and do the assignment as well. We added four Easy hard, very hard, something like that, try them as well, submit in the process section. And stay curious, keep experimenting until next time, happy prompting. 7. Refining Responses: The Art of Iterative Prompting in ChatGPT: Have you filed your conversation with AI weren't quite getting the result you wanted? Well, today we are diving into concept that might change how you think about working with JAG Pet. And the name of topic is mrollPs Iterative ferment. Now I'm going to let you in a little secret. What if I told you that the way you chat with an AI can have a huge impact on the quality of its response? Yep, we are not talking about just asking one and being done. There's something deeper going on. Something that's kind of like having an evolving conversation with a friend. I know you are interviewed. So state for that. So let me share a quick story. A few months ago, I was helping a friend write a letter to apply for a volunteer position at local Animal Shelter. We kept refining it, making small changes, like adding a little more about her experience with pets and adjusting a tone and sound a bit more compassionate. At a point, we looked back at our first drop and we couldn't believe how different it was from the final virgin. We didn't just make one change, we iterated, refined and shaped it over several exchanges, and you know what, she got the position. So what does this have to do with AI? Well, this same process of refining and shaping is a crucial when you interact with hi hippity. But here's where it gets interesting. Unlike regular connotiation, when you refine through iterative proms, the way the model response is different, so let's explore why. Now, just imagine you ask JAG PT to draft an email. You get a response, and it's not bad, but it's not exactly what you wanted, either. So you give it some feedback. Like, can you add more details about the project deadlines and alla? And the new version is closer to what you envisioned. But what if you start a brand new conversation and give it exact same instruction? You might end up getting a very different result. Why you ask Because interative reforment isn't just about stacking more instruction. It's about building on context of previous response. So do you remember in previous video, I was telling that Cha Jept has a memory? It remembers everything. And if you ask like a generic prompt, again, still, it will give you the output, and that output will resonate with your in vision. So if you didn't got that point in previous video, so we focus on this video because I'm explaining everything, like how memory works and what should be doing while adding the prompt and why we shouldn't always start our conversation brand new. So this means every response, every sentence you see on the screen become part of the large conversation. GPT doesn't just consider the latest prom. It's looking at everything that's been said before. Let's say, we are planning a vacation, and you ask ha JPT to search some itinaries. After the first suggestion, you ask, can you recommend activities that are more family friendly? The second response will be based not only on the latest request, but also on what you mentioned earlier, the destination, the travel dates, and extra. If you were to ask the same question in brand new conversation, it might not remember. That it's a family trip. This continuity is a power of iterative refinement. Now, here's where it gets a little deeper. In Ji GPT, every interaction is a part of one big ongoing prompt. That's right. Even though it feels like you are typing separate message, the model is actually seeing the whole entire thread. So if you start a new conversation each time, hi GPD teats it like a blank slate. But if you keep refining within the same thread, you are building a conversation history that shapes and inform each new response. So when you refine through conversation, the AA isn't just using the latest message. It's using all of its previous response too. That means its own outputs become part of the context of the future replies. Pretty good, right? So you are not just adding the layers with each prompt. You are creating dynamic, evolving version that lets Jagipti become more aligned with your needs. Now let's look at the big picture. When we refine iteratively, we are not just adding instruction. We are creating a sheer context. Think of it's like sculpting a statue or a block or marble with each refinement. We are lost starting over. We are sizzling away, slowly revealing the masterpiece inside. Now, if you were to start afresh with each Hammer strike, you had never get a beautiful statue, right? Similarly, if you keep switching conversation in ha GPT, you lose that sheer context. But by refining iteratively in the same thread, you are presuming that context and letting the AI understand more deeply. Okay, I hope you understand what I'm trying to explain it over here. If you're not, let me add another example. Imagine explaining your dream house to architect. You start by saying, I want two story home, and a few minutes later, you add Oh, and big garden would be nice? Now, instead of starting a new conversation every time, you add detail. You are building upon existing vision. Getting the architect or in our case, the Chachi pity a more complete picture. So what's the key takeaway here? Iterative refinement means evolving your conversation with Chat chipot rather than starting fresh each time. Each responses the AI gives become part of the conversation, influencing its next reply. And last takeaway is this process allows for a deeper understanding and more precise output, as you are not just giving instruction, but also building context. So understanding this can help you get more personalized, accurate response from hatchipet making your interaction more productive and fulfilling. So if you found this cell pool, please let me know. I love to hear how you use itty refinement in your conversation and also check out upcoming videos on how to handle negative outputs during atery refinement. So there's a photo video, and I will see you guys in the next one. Bio. 8. Writing With Depth: In-Context Learning Techniques in ChatGPT: Have you wondered how AI to like hat Jeopardy learns and understand from the information we provide? Imagine unlocking the secret to make your interaction with AI more smarter and intuitive. So today we are dive into a fascinating world of context learning, writing, and information density in hit Jeopardy. Trust me, by the end of this video, you will see AI in whole new light. On Ciner. I have spent years exploring how AI can enhance our everyday lives. So just the other day, I was trying to get haiti to help me draft a heartfelt birthday message for my grandmother. At first, the suggestion felt like a bit generic. But then I discovered the magic of incontext learning. Suddenly, the AI was crafting message that felt personal and genuine. That's when I realized how powerful understanding context can be. So what exactly is in incontext learning, and why should you care? In simple terms, it's how AI use the information you provide within a conversation to generate meaningful and relevant response. Whether you are a student, a professional or just curious, understanding this can transform how you interact with AI to like study pity. So in today's do, we will explore three key areas. First one is in context learning, how AI understands and uses the information you give it. Second one is effective writing with AI. Tips to make your AI generated content shine, and third one is information density, maximizing the value of the data you provide. Plus, we will share some surprising insight that might just blow your mind. Let's kick things off with in context learning. Think of it as a teaching AI by examples. When you provide specific information or example within your organization, hat JBT uses that to shape its response. For instance, imagine you are helping your friend a plan a surprise party. You might say, we are planning a garden theme surprise party with lots of colorful decoration and fresh flowers. Chat Jibty takes that context and suggests ideas that fit perfectly with your theme. By giving clear examples, you are guiding the AI to understand exactly what you're looking for. It's like setting the stage for the play. Providing the right cues ensure the performance sits all the right notes. Now next step is writing with AI. Whether you are drafting an email, creating a story or working on a report, ha JP can be invaluable tool, but how do we make sure the output is just right? Here's the simple tip. Be clear and specific with your instruction. Instead of saying write a story, you can try, write a short story about a young explorer scoring a hidden waterfall in the rainforest. This helps the AI generate content that's aligned with your vision. By providing detailed prompts, you are not only improving the quality of writing, but also ensuring its response with your intended audience. Now let's talk about information density. This is all about how much useful information you can pack into your forms. Like high information density means you are giving AI a lot to work with in a conscious way. Imagine you are assembling a person. The more pieces you provide, the clearer the picture becomes. Similarly, when you offer rich conscious detail, hag Bitty can produce more accurate and relevant response without overwhelming with you with unnecessary information. For example, instead of saying, tell me about space, you could ask, explain how Black Hole forms and their impact on surrounding galaxy. This focused approach helped the AI deliver the precise and valuable insights. So the other day, my sister Sapna was giving the reply to assignment, like you submit assignments, and she responds. She's the one who handles the back end of these courses, and she also edits the video. So she was asking me, like, trying to, like, there is one Asa met, and I'm trying to give some really personalized, like, a response. But she was not able to do this. But then I told you have to give context, as well as you have to add some personal tune to it, and also the tips we saw in this video. Then she tried it, and then she told me, now it works accurately. Now, let's recap what we have learned today. In context, learning allows AI to understand and utilize the information you provide within your conversation. Effective writing with AI involves giving clear and specific instruction to get the best result. Third one is information density is about providing reach, concise detail to maximize the value of your prompts. Understanding these concepts can significantly enhance your interaction with chat Jeopardy, making it powerful ally in your personal and professional endeavors. If you found this video helpful, please let me know. We love to hear from you. And stay informed next video where we will explore advanced tips on fine tune your air interaction even further. As always, I provided examples, check him, try it on your own, and try to add these tips in your prompts in daily prompts. And I guess there will be a summit to that project do that summit in the process. So until then keep exploring and harnessing the power pay. That's me signing off. Thanks watching, and I will see you guys in the next one is out. I 9. Using Persona Patterns for Unique Writing Styles in ChatGPT: Everyone, Chain. And it's freezing cold today, and that's why I made this room a little bit cozier for us. If I seem extra energetic today, because I just got from the gym, and I'm feeling super pen to make these videos. Well, today we're down to something that is going to change how we use hat JB. And it is called writing post on a pattern. And I'm not going to lie. This is one of my favorite prompt engineer technique. Because of this technique, I created lots of cool content on all the platforms. If you master this technique and there is prompt engineering technique, then you don't need to even learn prompt engineering. If you're in content creation, if you write blog post or any social meter that includes writing, then you don't need to learn prompt engineering. I will suggest you learn other topics as well, because it is only work for writing stuff because other techniques are helpful for other stuff, so you have to learn other techniques as well. You know, I remember the first time I tried to use ha GPT to help me draft an email for a job application. Let's just say it didn't quite reflected my professional tone. But then I discovered this awesome trick and wow what a difference it made. So that time, I just graduated from my engineering, computer science engineering, and I was trying to get a job. I was also doing content creation on all the ATC platforms, but I was just trying like, let's get a job because I was not serious about ATC. Now I do this full time. So that time Cha GP just came out, and I was just trying, like, let's see, can Cha JBT create a job application for me? That time I was just learning prompt engineering. I was just experimenting with hat GPD. So while experimenting, I got this technique, the writing pos on a pattern. And that time I didn't even know there is a name for this technique. Like now I remember there are 60 40 techniques. They unnamed. But when I recall it, it is like I do this on daily basis, and there are several different techniques and named and this technique named this. So I get super confused these are just my daily habits. This is how I write prompts, and there is a one field called prompt engineering. Then I got to prom engining, and then I started teaching prompt engineering. Just got first move advent. And that, I was just experimenting with Chad GPD. So that's why I decided let's teach engining. Because when I was in engineering, I had honor subject as my AI machine learning data science was my honor subject, and I got outstanding grade in it. So that's why I thought, I'm grading this, so let's teach engineering or a related stuff. So why should you care about writing person a pattern? Well, if you ever wanted Charge Berry to write more like you, or if you have been first with a generic response, this is the solution you have been waiting for. In this video, we will cover what the writing person a pattern is, how to use it, and why it is so powerful. Get ready to have your mind blown. Writing post on a pattern is like giving ha Gipeti a crash course in being you. It's a special way of formatting your prompts that helps Cha Jipiti understand your writing style and voice. Think of it like this. Imagine you are teaching a parrot to talk. Instead of just saying hello over and over, you give it an example of how you speak, the words you use and even the tone of your voice, that's what writing post on a pattern does for ha Jipity. When I was in school, there was a story about parrot. Like, there was one person who wants to do experiment. So he bought two parrot, and he gave first parrot to, like, a family who was, like, decent. They don't usually fight that much. And in second family, and he gave the second parrot to a family who use a bad language a lot. So after year, he came to meet those parrot. First, he went to a family who fights a lot. And when he was, like, entering in a house, the parrot was outside on the porch, and he started speaking bad language to him. And then he not even went inside the house. He just moved out and went to another home. The family was decent. So the parrot was, like, welcoming him in, like, more warm way. So those parrot just picked up like how the environmentter was. They got to know about the tone of those family and they just picked it up and started speaking like them. Like this is what we do in prompt engineering. I hope you got the example. If you don't got that example, let's take another example. Let's say you want to write a blog post about your favorite hobby, like knitting. Instead of just asking Chi Jib to write a blog post about knitting, to show how you talk about knitting. Now let's get into nitty gritty of how you use this pattern. It's simpler than you might think. The writing person a pattern has two main bads. First one is instruction on what and how to write, and second one is example of your writing style. So here's the template. Like first introduction in that ad, what do you want to chat GBD to write and how? And in example, past some writing material. Like let's say if you have content from a blog or maybe from social media post, copy that and add it into example. Now let's go back to our knitting blog example. This example, I wrote this kind of prompt. You can also mention, instruction and example if you want. And then add what you want, like at the end, I applied, like, now write a blog post introduction about the joy of baking in style of example above. So first, I gave the instruction, write a blog post. Then I add example like how I talk, maybe something like that. And don't worry. I don't talk like this or I don't write like this. I just took a example from the Internet, and then I told Chat Jepity write a blog post, introducing about the joy of baking in style example above. So let's give this to ha Jepity. Okay, so we got dopood and output is like this. Hey there, fellow baking lovers. Today, I'm super excited to share one of my all time favorite ways to relax and have fun baking. And there is just something so satisfying about mixing ingredients and watching dough raise and, of course, the amazing aroma that fills the kitchen. Whether a season baker or getting started, stick around too, we whisk kneaded and sprinkle our way into delicious world of cakes and cookies and everything sweet. So as you can see, if you read the example, it is very similar to this example, the output. So this is just a basic example. And if you master this technique, you can I will tell you how I use this. So, listen, if there is one YouTube video or there is a one video and I like the lot. So what I do is that I took that transcript and I passed it to Cha JPEt and I ask JAG Put or I tell JAG Pet, like, please tell me, like what kind of tone in that, how he's teaching, and what kind of tone he is using something like that. And I also add reverse prompt engineing technique in that and I also say ha JP reverse engineering this script and write me a prompt. Doesn't always work, but you have to add irritative refinement to it, and then you got your prompt. That's so simple. That's why I was saying, this technique helped me a lot in making content creation. So as you can see, by giving ChagpD an example of how you write, it can mimic your style much more accurately. Now, here's where things get really exciting. The writing post on a pattern isn't just about making Chi Gipe sound like. It's about taking control of AI and making it work for you. When you use this pattern, you are not just getting better writing. You are teaching the AI to understand your unique perspective, your experiences, and your way of thinking. This means you can use Ja GPT to brainstorm ideas, draft articles, and even help with creative writing projects. Imagine you are a teacher trying to write engaging lesson plan. By using this writing post on a pattern, you could show chat GP examples of your best lesson and then ask you to help brainstorm new ideas in your style. There is a lesson plan that sounds like you wrote them, but with fresh new ideas to keep your student excited. So there you have folks. The writing post on a pattern is your secret weapon of getting the most out of JaGiPty. So let's recap what we have learned. The writing person a pattern helps ha GPI write more like you. It consists of instruction and examples of your writing. By using this pattern, you can personalize AI airports and make them more authentic. Remember, AI tool like HGPD are here to help us, not to replace us. The writing persona pattern is a fantastic way to harness the power of AI while keeping your unique voice and perspective. You want to learn more, try experimenting with the writing person pattern yourself. Start with something simple like writing a social media post and see how it goes. And don't forget to share your experiment with me. We would love to hear how it works for you. If you want to practice this method, there are lots of examples add in like four or five examples I added. Copy them, try it on your own, and you can add your own data into it, add examples into it. A adds to that summit project section. I hope you found this technique helpful, and I hope it will become one of your favorite prompt engineering technique. So that's about what you do, and I'll see you in the next one. Up. 10. Choosing the Right Examples for In-Context Learning in ChatGPT: Have you ever asked Cha GPI for the help and thought, This doesn't exactly sound what I expected. Well, there is a reason for that, and it all comes down to example you provided to Cha JPE. Well, today we're diving into why we examples are critical for in context learning. Let me tell you about a time when I tried to use Cha J Pi to friendly birthday message. And instead, it sounded like I was giving a formal speech at University. That's when I realized the power of examples. Today, we are going to break down why the example we choose are so important for guiding AI response, and how you can use this knowledge to get better results and a few fun tips along the way. So are you ready? Let's get started? First of, let's talk about why example selection is crucial. Think of ha JP like a student in a class. If you only show the student how to solve one type of math problem like let's say addition, they will only be able to handle additional question. The same things happen when you que ha GPI a specific example. It learns from the examples to answer similar questions. Imagine I ask GPT to write a letter of recommendation for students using an example that are too formal, like a contact letter. When I use that contract style writing, it wouldn't certainly know how to sound warm or personal, just like asking someone who only knows how to add to solve a substraction problem. Now you might be asking Chetan, then what is the solution? The solution is, choose the right example for the right task. Now, let's look at what happens when you provide the mismatch example. Let's say I have been using HGPT to help me write professional reports. But one day, I needed it to help me write a fun invitation for Kills birthday party. If I field it only former reports, guess what? That birthday invite will end up sounding like a work memo. Like, please join us for the celebratory event making the birthday anniversary for minor individuals. Yikes, right? That's not what you want, right? The point is examples actors guide, if you're teaching it how to write one way and don't expect it to switch gears without new instruction. Now, here's the exciting part. How do you actually choose the right example? It's simple. Think about what kind of task you want Cha Jeb to perform. Like, do you want it to write a friendly email? Use friendly conversation example. Need help in writing speech, provide with a speech example. Look at this. Here's an example for professional versus Schedule one. If I want Cha Jeb to write something casual, I need to show it Casual one, not form of one. Just like you wouldn't use the recipe for the cake to make spaghetti. You wouldn't use the wrong writing style to get certain result. This is where the magic of in context learning comes in. So if you're still confused, like from where we get in context for, like, writing email or maybe writing for the message. So what you can do is that, I always do this. And my lots of students ask the same question. Like, let's say there is a scenario, like, I want to write email. But probably with that email is, I want to write in very, like, a friendly to so I told them like Asta epity like, start a new chat and ask Ta Jept Here is the problem. I want to write mail or message for my friend and who is having this kind of problem or you want to ask something like that. Tell exactly what you're facing currently, and then ask Ta Gpity like give me context or example. So I can provide in email to get the result what I want. So let's now recap this. So what did we learn today? When you are using to chat betting the example you choose make all the difference. They shape how it responds and help it get the result you need. Whether you are writing a fun birthday invitation or a serious report, matching your examples to your task is a key. Now it's your turn. Let me know how you are going to use this information to write a email or text or how you're going to use this in your daily life. And if you are curious about more tips like this, trust me, you won't want to miss our next video where we will dive even deeper into making AI works for you. So that's all about why example selection is crucial for GPD in context learning. So that's what video. I hope you learned and you got the point why we provide examples. So that's what video, and I will see you in the next one. 11. Customizing Prompts for Personal Preferences in ChatGPT: Have you wished you could get perfectly the right answer from AI every time. What if you could train the system to align perfectly with your own preferences, making it more useful for whatever task you are working on. By the end of this video, you will have the tool to do just that. I remember when I first started using AI for everyday task, like for email, for writing post, for creating script, just assume for everything. So I had type in a or a request, and sometimes the response was spot on, but other times it felled little off. I had think, how can I get it closer to what I really want? And then I discovered the method that completely transformed the way I interact with AI, which I'm excited to share with you today. Today, we are diving into preference driven refinement of prompts, a method that allows you to teach AI how to respond more closely to your style and needs. It's all about refining and guiding the system to meet your preference, making the AI work better for you. We will break this process down into three simple steps. I will share example, stories, and by the end, you will be able to apply this method to your own interaction with ha Jepidy or any other air tool like ad Baxlty Gemini. So let's get started. The first thing you need to do is create a basic initial prompt. This is the starting point for your interaction with AI. The prompt doesn't need to be perfect right away. It's all about getting process started. So let's take a very simple example. Let's say you want the AI to help you write a birthday card for a friend. You might just start typing like this, write a birthday message for my best friend. So let's give this to habit. So as you can see, had Gib generates a nice message, but maybe it says something like another year of laughter, adventure and memories. So this is not my style, and I don't like these kind of, like, words, laughter, adventure and memories. My writing style is more friendly and warm kind of thing. But this is like, too excited, something like that. So basically, this is not my style. It is okay, but we are getting started. The first from allows the AI to generate an initial response. From here, we will begin the response or pfying the output. Now that we have a first response, it's time to move on to refine it using preference driven feedback. The key to this method is teaching the AI what you like and don't like about response. For example, if birdie messages sound or the top, you can tell the AI what part you like the most and what you didn't. So in this case, what we can do is that we can ask at Jeopardy for write a body message for my friend again. And we can change the tone like write in friendly tone or write in warm tone. We can ask it two, three times. And then what we are going to do is that, we will collect the part we like in this output. So in this part I like, you are the most incredible friend anyone could I like that line. So I will copy that and I will also copy the output from other three prompts. And I will say, This is what I like in this prawn. And I will also mention, like, another year of laughter, adventure, and memories. I didn't like this part. I will also collect those from other three outputs, and I will also add that in prawn. I didn't like this output. And I will also say to Chad Jeopardy, learn from this point and at the end craft me a final bird message. So let's do that. Okay, let me tell you one trek. I don't use this usually, but there is one ik. So as you can see, I wrote, write a birthday message for my best friend in a warm tone. But if, let's say, I'm not getting any idea, in that case, you just can add space, space, just add space and just give it to Cha Jept. Cha GPT still will produce new output. It will not produce exactly the same output like this. It will produce totally different output. So as you can see, the first line wishing you the first happy birthday. But in this output, it's finally your birthday, and I'm so excited to celebrate you. So now we have three outputs. Let's collect them, and let's combine then Iot prompt. And let's tell Ja Jept This is what I like and also tell ha GPT. This I don't like. And then we will ask write like birthday message for my best friend and find a version. It has to find a version. Okay, I'm going to do that. Okay, I collected everything from these three outputs, like what I like and I don't like. So as you can see here, I told Chap, I like the part we mentioned. My friend is caring about thoughtful and keep the style. I also mentioned three points four points from each output, like first one, this one, second one, this one, I also mentioned the numbering over here. And then I also mentioned, I don't like the fridges. And I also extra from the four outputs. And at the end, I wrote, No, finally, you know what I like and don't. Now, write me bidday message for my friend. So let's give this So if I read the final output, I'm so excited to celebrate with you. You are the most incredible friend anyone could ask for, and I feel so lucky to have in my life you bring such light, kindness and everyone around you. I hope today, I'm feeling with you all the love and happiness that you deserve. I wish you nothing but goodness, vibe, and lots of laughter ahead. So as you can see, as compared to these three, this is one of the best output I will say, because this output was a little too much for me, and this one is, like, perfectly hit the spot. So I like this one. So as you can see, this is where magic happens. By repeating this feedback process, you can create a prompt that constantly genders response aligned with your personal style. Imagine being able to customize not only birthday message, but email, social media post, or even study guides, all trailer to your voice. This method of refining prompts allows you to take control of EI or puts in powerful way. Okay, not only you can use this method, you can combine this method by adding your own message, like we saw in previous videos, like we used to give example of our own writing style. And what we can do is that, so we will write like this, I like this one, and like this one, and then we will write now write me message. But you can copy this example of mine. This is how I write and then write me output or write me message. Then it will write on your own. So by doing this, we are combining two, three prompts in a one to get the output we need. Okay, let me tell you one real story that happened. I had a student who wanted to write a motivational speech for community event. First drop was good, but it felt too formal and lacked the personal touch she wanted. By using preference divine refinement pattern, she gradually shaped the AI response until the speech was heartfelt and impactful, just the way she envisoned. To recap this topic, here are the key steps in preference devi refinements. First only start with a simple prompt, evaluate the output and identify what you like and you don't like. At the end, feed that field back back into AI by providing examples of both. This process not only saves you time, but also gives you more control over the AI responses, making it useful tool in your personal and professional life. If you are eager to learn more, I encourage you to experiment this method in different solution. And don't forget to tell me how you are going to use this method in your daily life. Okay, go to resources and check other examples. There is also examples as well as output, like how to write prompt and do that, as well as I also give assigned projects, check them out, try to solve them and smdnty projects section. So that's it. I hope you understand this technique, and that's photo today. Now I will see you guys in the next one. Miss out. 12. Five Creative Ways to Tackle Prompt Challenges in ChatGPT: Have you faced a problem and you just didn't know where to start? Maybe you tried a few things, but nothing quite work. Well, what if I told you there is a way to get not only one solution, but to get multiple solutions for your problem. And it will happen within a matter of seconds. That's right. Today we are diving into five ways to solve problem using hi Gi petty. That will completely change the way how you approach problem solving. By the end of this video, you will know how to tap into full power of AI to explore creative solution in ways you never imagined. By the way, I got this money plant, and there is also one other plant, and there is another plant that is also many plant. This is snake plant. Let me show you. This neck plant. I just kept it to get some oxygen because when I lock this room, there is a very less oxygen. So I thought let's bring some plants. There are other plants as well in hall. There are 34 money plants and other plants as well. I like plants a lot, so that's why I keep them. They are like pets to me, so I keep them. Okay, now let's start with the topic. I remember when I started using AI, I thought it just going to give me one answer, right? But then I realized it can do much more. You can get five, ten, or even ways to tackle challenge. And compare which one works best for you. It's like having a brainstorming session with an endless number of ideas on the table. And I can't wait to show you how easy and effective it can be. So why does this matter? Whether you are planning a family event, troubleshooting or tag issue or deciding how to organize your closet. Knowing how to find multiple approach to solve a problem gives you more control and options. And which debit you can make smarter choice faster. In this video, you will explore simple ways you can ask JAG PR for multiple solutions. How to weigh the pros and cons of each one and fun bonus. And some examples of unique problem solving methods that will surprise you. Now, the first and simplest way to solve a problem using hi GPT is by asking for option. Instead of just requesting a one solution, why not ask for several? Chi JBT is great at giving you variety, whether you are stuck on how to arrange a garden or deciding what activities to plan for kids birthday, asking for options opens your mind to a different possibilities. Let's say, you want to plan a family movie night, but you are not sure which movie to choose. Instead of asking Cha Ji PR for just one suggestion, you can try prompt like this. And let me tell you one thing. Like when I was in engineering, there was a group of people, and they created this project like they created movie recommendation system and if, let's say, that in lognamT there was a time where lots of people used to watch lots of movie. Like, I remember I was the one who used to watch two, three movies in a day. And at the end, there was a point like I saw lots of movies, and now I want to watch something else. And I used to go on Google, and I used to try like I used to search for different genres. I was like, I watched everything. Now, like, tell me what should I watch. Then I discovered German, then I discovered French Cinema, and I started watching those, and I also enjoyed them. That time Chat Jebed was not there. But on that problem, my classmate created a project, like movie recommendation system, and it took them like months to create that project. But as you can see, Chat Ji Bit is really simple. We just have to write a prompt, like give me five different movie genres, a suggestion on one top movie to pick from each genre. Then compare and contrast which genre best fit from family movie night and in this prompt, we can also add examples. We already saw examples. If we add examples, we get more accurate results. But in this case, we are not adding any examples because we are learning how to solve a problem in a different phase. Okay, so we got some suggestion, and let me tell you, if I read the output, I watched every movie, every single movie from the list, like I watched Indiana Jones, as well as the Toy Story. It is classic. Everyone watched it. The Incredibles At is one of my favorite movie. And second Potter is also good. And I guess in India, the dubbing of that main character was ShaouKan. Shar can gave the voice for that main character. And we also have Harry Potter. It is also one of my favorite movies. Then I heard the greatest show a lot, and I also saw lots of video, but I didn't watch it yet, but I will one day, I will watch it. But I know what's the plot and all that. And then we got compare and contrast for our movie night. And, okay, it is saying it is comparing all the movies from the list, and then telling us you should watch Toy Story because it's family. It didn't mention family, if you mentioned, friends, then it will say watch Indiana Jones, something like that. So that's how amazing Chad Betty is. So as you can see, you had a problem like which movies to watch, but you didn't know which movie to watch. And still Jadi Betty gave you this list. And it not only gave you the list, it also said, um, if you're watching with family, then what's pro story. So this is the creative way to solve the problem. Now that you have got the multiple solution, what next? The real magic happens when you compare and contrast them. Cha Gibt can break down the benefit and challenges of each option, helping you make more informed decision. But we was lucky in this case. Cha Jibty compared and contrast on its own, so we don't have to do it. But the day by day ha JBR is getting smarter, and I think in our problem, we must have wrote the comparing and contrast. That's why it is saying that's why it compared and contrast for us. If you don't write it and if you just got the list, then you can write a separate problem, comparing contrast and tell me which is the best movie to watch with family. So from this, chat Jeoparty break down the benefits and challenges of each helping you make a more informed decision. You are planning a weekend getaway and are sure where to go. So you could ask S Jeopardy like this. And by the way, let me tell you one thing. I like to travel a lot. That's why I took lots of traveling example in this course. And I will also add some photos of my recent traveling. Or you can also check my ISGAm. They are just filled with my travel photos, Acutrak and I like to go into Mountains and nature. So if you are interested in that, you can watch them as well. So for that, I wrote this prom text. So just five weekend getaway destination within three hour drive from Pune. So this is where I live in India, and then compare them based on weather, cost, and main activities at each location. So let's give this and let's see. Okay, Look, so first one, we got Lonola and it's kind of like one of the popular places in Punit It was a hill station, and my college was in the city, and I traveled a lot over there, and the Las second one. I also went there recently like five months ago. And, okay, I also went here also. You know, I went to all the location the hadibi Dimension. And this is one of the, like, amazing places to visit, um, where I live. Okay, ha Jeb is really smart. It also gave the table format, like, and it also compared weather cost and activities. So Lonolas really close. Lavas is also close. Marvel is quite far away from my location. And the Martin and I went like, two or one month ago. I will also show you the ***** you can, uh, watch on the screen. Okay. It also compared with the cost activities and at the end it's saying Lonola. Lonola is, like, really popular location in Pune. If you ask anyone, where should go, they will say, let's go to Lonola because it is, like, really easy to go there, and weather is also nice. So as you can see, by using this method, suddenly, we have got clear view of which option suits your needs and best without having to spend hours of doing research. Now, here's where things get really interesting. By exploring different perspective, you start thinking about problem in a way you never thought possible. This can lead to unexpected and often better solution. Let's say you are in a tricky situation, you have invited friends and you want to make something special, but you don't have much time. So for this, I wrote prompt like this. Like I have some ingredients, and I wrote, like, here are some ingredients I have. I wrote some, like, vegetables. You can read it. And then at then I like ask Jeopardy. Suggest three quicker dinner ideas using this ingredient and compare their difficulty level on time to prepare. Okay, I have to say the output is, like, really amazing and it still give us the this sable. This is what I like about ha Gibeti. Okay. So first one is vegetable stir fry with rice and maybe Keno What is that? Kuna I don't know how to pronounce it. Then we have vegetable wrap, then we have vegetable rice and Kiano bowl. Okay, we got some easy to make recipe. It also mentioned the difficulty level over here. Difficulty is easy. And at the end it told us like make vegetables stir fry. Okay, this is good. But if let's say if you don't mention any recipe and if you just mention your region like from India or maybe from Haasta, maybe from Italy, it will suggest the easiest option to make if you don't provide the ingredients. So one of the idea might turn out better than what you had in mind, save your time and impressing your friends or guest. So to wrap this herb using CharJP to ask multiple solution to a problem is a game changer. You can gather options, compare the pros and cons, and explore new perspective that lead to better outcomes. Whether you are planning an event, solving take issue or just looking at new ideas, this method can help you unlock the full potential of AI and make smarter quicker decision. I hope you understand how this method work and try your own methods, like try to add your own problems and tell chat petty give me different ways to solve this and also compare and contrast and try to just experiment try to modify the prompt. And I also added examples in resources, check that out, also added assignment, do that after doing that, submit enterprise section. So that's what Zo, and I hope you understand everything what trying to what we are trying to do here. So this is what DZ doo, and I'll see you when the next one is out. Or signing. 13. Different Approaches to AI Generation in ChatGPT: Have you wondered how you can use AI to not just get your answer quickly, but to make them better? Today, we are going to learn a simple trick that will completely change the way how you generate AI and Cha JEPD, and it is going to spark your creativity like near before. You know, I remember when I first started using Chichi PT, I thought it was all about speed and how fast can I get something done. But over time, I realize something crucial. It's not just about getting things done quicker. It's about improving the quality of what we create, and trust me, you don't need to be a tech expert to do this. Today we are going to talk about explotingGeneration, a simple technique that helps you unlock had Jept's true potential by generating multiple ideas, options, or versions. This way, you will create something far better than you started with, and it won't take much more time. In previous, we kind of use this technique like we used to say hat GPT, write me these kind of version of birthday messages, and then add then views to tell JatJPT is right, This is wrong. This is I like, this is what I like. And then ad then used to us, Craft me birthday message on what I like and what I don't like. We will go through how to use this technique step by step. With examples, that will really open your eyes to what's possible. By the end, you will see why this approach can lead to more thoughtful emails, creative solution, or maybe even spark a few wow moments. Okay, let's start with basic use of chair Jeopardy. Let's say you want to write birthday message, you could ask it to write one, and you get a pretty good result. But here's the treat. Don't stop there. Ask you to write three different version. You will be surprised at how each one has significantly different tone or style. Maybe one is funny. Second one is formal and last one is casual. Now, instead of just picking the first one, you can mix and match. Like take a funny greetings and add touch of formality. For example, let's take same example we saw in previous video, like write M B message from my grandmother. But in this, I wrote, Make It heartfelt and personal. Let's give this to Chile Deputy. Okay, so we got output for the heartfelt and personal. Now, let's ask, like, write me second version of Bird message of my grandmother, the same message, but write it in light hearted and humorous. And for third message, I'm going to write me in formal sculTne. Now, we got three versions of birthday message for grandmother. Now what we can do is that. Like we can combine them together, and then we can craft the final draft. Now, let's dive deeper. Let's say you are working on creative projects like designing a Community poster. You could ask GBD for ideas, but why not ask for multiple ideas like a five or even ten different tag lines? This lets you compare, combine and refine them until you get something that stands out. Imagine I need a tagline for my school project, so I can suggest five different tag lines for school event focusing on learning and adventure. So let's give this. We will obviously get five point, like explore, learn, sore and beyond the classroom. Now, let's ask something different, not different, but we will change something in this prom. Like I wrote. Now, give me five different tagline for school event that empathize and discovery and new horizons. Okay, so we got the output from these two tag lines. We got 55, four each. Now, let's ask Chagbt combine the elements. Let's assume this hagibt will take elements from this and this, and at the end, it will create a new tag line for us. Now, this is how this exploding generation work. And I know this line is not that powerful, but you can try multiple times. You can use iterative refinement and then then you will get the line. You will be proud of that. Or you can use the line from this tone, or you can add as an example, add this output as an example, until JDP, learn from them and at the end, combine them or mix match them and create me or write me new tag line. Now you might be thinking like, how to choose from this output or how to combine them to better and why we should do that. Now let's decode. Like here's where things get more exciting. When you have multiple version of something, your brain gets to work. You are not just clicking Send on the first thing that pops up. You are engaging with your material, making decision. This process sparks creativity, whether you are crafting a heartfelt message or planning an event. Think about writing an imported email. Instead of just generating one version, ask for several and compare them. Why does one feel more professional? Why does another sound more friendly? And at the end, you start analyzing the small details, tone, and word of choice, and before you know it, you have crafted the perfect message, all while exercising your creative muscles. So what did we learn today? By exploting generation, you can use cha chi petty not just for getting things done faster, but to create better, more thoughtful content, whether it's writing an email, designing a poster, or creating a message. Generating multiple options help you make informed creative decision. The beauty of this technique is that it doesn't take much more time, but the impact is huge. You get to compare ideas, blend the best of different suggestion and ultimately create something that's truly your own. I know in this video, the output was not that amazing. If you try any prompt, not just stick to first output, try multiple versions of them, then compare and contrast, then mix match them, then tell Charpy this is what you like, this is don't you like, then at the end, draft a final thing or final output for that. Now that you know how to exploit generation, give it a try, start with something simple, maybe bird a message or an email and see how generating multiple versin can improve the final product. You will be amazed at the result. You found this video helpful, please let me know and try to experiment this technique. And let's say, I will give you a task like write a message for your friend and try three different version of that message, and at the end, combine three of them or combine two of them and craft final version for that. If you don't like final version, you can edit your prompt, you did that output, and try to create another output. Do it until you get satisfied and then send that message to your friend or family member. Resources, I added examples, try them, also added assemment, do that, and after completing it, semito project section. That's what data will do and I'll see you as in the next one out. 14. Creating Metrics for Evaluating AI Responses in ChatGPT: Imagine if you could take any product idea or project and instantly know the best way to measure it success, if I told you, there is a way and you can do this in very simple steps using AI. Today we are unlocking the secret, and by the end of this video, you will see assessment metrics in whole new light. I remember when I was working on planning a community event, we had so many ideas about how to make it successful, but we had no clear way to measure which one would work best. It was overwhelming. But then I started thinking, What if I could break this down with clear metrics? That's where I discovered how hat Jeopardy could help me with structure assessment, and it was a game changer. So what exactly are assessment metrics? And why do they matter? Whether you are depending on the best way to run an event, choosing between different strategy for your project or even figuring out how to approach a personal goal, having a set of criteria or metrics to compare your opinion or options can help you make informed and effective decision. When you are starting out, the first thing to understand is that generating assessment matrix doesn't have to be complicated. Let's say you are trying to decide how to spend your weekend. You could measure the options by things like fun level, cost, or travel time, and also energy requirement. Chat JEPD can help you organize these thoughts and even suggest additional metrics you haven't considered. For instance, you could say, give me five matters to evaluate different weekend activities, but don't list the activities themselves. So as you can see, chat Gaby suggests enjoyment level, social interaction, energy boost, cost effectiveness and learning and kill growth. So with just few line, you have got a structure way to compare going on a hike, sit a museum or staying home for a movie Myathon. So as you can see, with just few lines, we have got the structure way to compare it going on. Now that you have some basic metrics, let's take it a step further. What if you are planning a bigger project? Like organizing a family reunion? In this case, you might need more detailed criteria to assess different venue or catering options. And here's where Chargabity really shines. You can ask you to generate metrics based on how complex tasks need, like sustainability for larger groups, ease of access, or even menu variety. Let's say you decided between three venues. For this example, you can ask prompt like this. So I wrote, send me five matrix to evaluate different venue for family reunion. Focus on things like sorry, suitability for large group accessibility and atmosphere. So changeability suggesors group capacity, accessibility, atmosphere, and amenities and activities, and also the catering and food options. So this help you compare each venue side by side, making a decision clearer and more informed. Now let's understand the third part of this video. That is advanced metric impactful decision. Here's where it gets really interesting. Imagine you are working on a project that could change your career or business, and you need a super detailed evaluation. You can ask Cha JP to not only generate metrics, but also help you rank or prioritize them based on what's most important to you. Let's take in completely different scenario. Let's say you're deciding on a new business strategy. You could have metrics like this. Prod me five metrics to e a different business strategy. I focus on cost efficiency, then customer satisfaction impact and implementation time. So as you can see, Chat GPT is helping us break down this even further. Like, first, we got cost efficiency, then impact on customer satisfaction. We also got the output. Like, how does this strategy affect your customer happiness? Also consider metics like feedback, fedex score, repeat business, and also improve loyalty. And it also gave us the tip like, will it elevate the customer experience or just be nice to have feature? If you want, you can ask to organize all this into kat table or chart to make comparison even easier. To summarize assessment metrics are your roadmap to making smarter decisions, whether you are choosing weekend plan or making life changing business modes. Chi JEPD simplifies the process by helping you generate, customize and organize these metrics. So you can focus on what matters most. This technique not only saves time but also adds clarity to your decision making process, empowering you to make choices that align with your goals and values. If you want to try this out for yourself, start by thinking of a decision you need to make. Ask GPT to generate metrics for evaluating your options and see how much easier the process becomes. Okay, I hope you understand what we are trying to do in this video and try to take any problem and take any problem, and on that, apply this assessment matrix prompt engining technique. So that's what we do, and I will see you in the next one. So. 15. Using Automated Search for Prompt Improvement in ChatGPT: Imagine being able to find perfect solution to a problem without spending hours of searching manually. What if if I told you Jaipt could do that automatically. But no matter how hard I tried, I kept missing little details. Then I discovered how ha JEPD could generate multiple solution, and suddenly everything clicked. That's when I realized how powerful automated search could be for everyone, not just for tech experts. Today, we are diving into automatic search in Ja JP, a simple but effective way to get multiple answer, a solution to a problem. And then quickly sort through them to find the best one. Even if you are not programmer or tech savvy person, knowing how this works can help you think differently about problem solving. In this video, I will explain what automatic search means and how JDBty can do it and how you can use it in everyday situation. Like finding the best solution to a tricky puzzle or even sorting through options when you are deciding something important. And we will also explore some mind blowing examples, so stick around. Now, let's understand what is automatic search. Automatic search simply means using AI like ha gibt, Cloud or perspexility, to create lots of solution to a problem, and then automatically testing or comparing them to find the best one. Think of it like having a group of people helping you bentorm ideas, but only faster. For example, let's say you are trying to plan a family trip, but you can't decide on best destination. With automatic search, you can ask Cha Jept to suggest a bunch of travel spots. Then based on your preference like budget, weather, or activities, you can rank these places. Then Cha Jept will generate multiple options for you, and you can automatically check which one fits your criteria the best. So for this example, I wrote prompt like this. Suggest me five vacation destination that would be great for family of four it a budget of $2,000 and mild weather and kid friendly activity, rank them based on affordability and all activities. Okay, so we got the output, but don't focus on the output. The main thing was the prompt. I just wanted to show you how this method work and how you're supposed to ride to get accurate result. And we also got some amazing places to visit under $2,000, so we can check them out as well. But I live in India, and I never visited the kind of these places, but I hope one day I will, and that time I will ask hagibt and I will use same technique like automatic search. Now let's look at more advanced ways to use automatic search. Suppose you are organizing a community event and need to decide on the best date, location, and activities. You could tell ha GPD the general idea you have, and it could generate several schedule and plan for you. But here where it gets interesting. Hi GPD can also help you rank this option based on factors like availability, cloud reference, or even weather forecast. Think of this like planning a birthday party. You could ask ha Gibt to come up with different party themes, venues and activities. Then you could give it the criteria like which theme is most cost effective or which venue has the most availability. Cha JBT will sort through the idea and help you find the best combination without you having check each options manually. So for this example, I wrote this kind of prompt like a general three birthday party, three, 14-year-old boy, including a suggestion for venue decoration activities. And then I wrote rank them based on the cost and venue availability. Okay. So as you can see, automated search is, like, really great. So we just asked like, what will be the cost and venue availability. So it also mentioned like DIY do it yourself and affordable. And what the birthday theme is superhero training cap. And we can have this party in Bad and Loc Park. So it will be affordable. And we also have second one is Adventure quest. Okay. This can be a bit pricey. And third one is Va reality Arcade game party. This is like it give stars like three stars, and this is expensive. Recently, I also experienced R, and I have to say it is expensive because having that huge setup or maybe ordering that setup, it is expensive than the Bhad party. So it depends on if you're rich. If you're rich, then you can totally aware or if you don't want to spend that amount of money, and then you can have a Bhad party. Like the theme is superhero training camp. Now, here's where automatic search becomes even more valuable. It's especially useful when the solution is tough to generate, but easier to check. For example, let's say you are entering a cooking competition and you need to come up with a unique recipe, you could ask JagiBT to generate several recipe ideas, and then oro taste tester can easily decide which one tastes best. So for example, picture this. You are designing a custom cake for celebration. You could describe your basic idea to ha Ji pit like flavor of combination and design theme and it could generate several creativity ideas for you. Instead of baking each time, you could quickly scan through the suggestion, choose the one that sounds most exciting or unique, and suddenly you have the perfect recipe without wasting time or ingredients. So for this example, I wrote this kind of pro, like suggest four creative cake design for 50 50th wedding anniversary included flavor of combination, decoration idea, and rank the design based on the uniqueness and complexity. Okay, so we got the output, and as you can see in output, we got flavor, decoration, the ranking, the uniqueness and complexity. So this was so this was the main purpose of the prompt. So as you can see, by using this method, we don't have to search a lot on Google or maybe on hat GPity. We can just write prompt like this, and it gave us the, like, really amazing output. And we can just, like, skim through this and we can choose any of this recipe, and we can tell had GipetinGive me the recipe, and we can work on those recipe to make cake. Automatic search and Cha chipoti help you quickly generate lots of options and trace them to find the best one. Whether you are planning an event, solving a puzzle or even coming up with creative ideas, it's like having an entire brainstorming team in your pocket. No matter what kind of challenge you are facing, using ha GPD for automatic search can save time and effort. While giving you the better range of solution is not just about tickets. Anyone can use it to find the best outcomes in everyday lives. If you want to give it a try, start by asking Cha JEPD for multiple ideas or solutions to something you are working on. You will be amazed at how quickly you can find the best options. So in resources, I added examples, the examples I use in this video, as well as some add on examples, copy them, try them out, try to experiment with them, also added assignment, do that. And this is what it, and I will see you well in the next one. 16. The Essential Parts of a Good Prompt in ChatGPT: Have you wished you could give better instruction to machine, like how you would give to your friends or family members and get exactly the results you wanted? Well, today we are talking about something that will change the way how you communicate with AI. That is the five components of the prompts. I remember the first time I tried to get an AI to help me write a simple email. I typed out what I thought were clear instruction, but what came back looked nothing like what I had in my mind. That's when I realized it wasn't the AI's fault. I needed to learn how to give a better prompt. So in this video, we are going to break down the five key components of the prompts. These are like building blocks that help ha Gipe understand exactly what we want. Whether you are chatting with AI to write story, answer a question, or solve a problem, knowing these components can make huge difference. And trust me, by the end of this video, you will able to master the art of prompts and get judge Body work like your ideal assistant. Now, let's start with the first and most obvious component. That is instruction. Think about when you ask someone to help you with something. You don't just say, do it. You give clear instruction, like, please bring me a glass of water. Same goes for prawns in ha Jepity. We need to tell it what we want. For example, imagine you are asking had Jept to help write a birthday card for a friend. Instead of just saying, write a card, you want to say, write a fun two paragraph birthday message that includes joke. Now let's imagine you have a younger nephew and you ask him to clean up his toys. If you don't specify which toys or where to put them, you might find all the toys showed under the couch. So as you can see, clear instruction makes all the difference. Now next is information. When you are working with an AI, it's like working with someone new on the job. They don't know everything yet. You need to give them the right facts to get the job done. Let's say you want Ja Gipe to help you plan a small gathering. If you don't tell it how many people are coming or what kind of food your guests like, it might suggest five course meal for 20 people, even though you are only having a casual dinner for four. The more specific information you provide, the better the results. Now here's where things get even more powerful examples, just like showing a friend how you fold a paper airplane or bake a cake. Example helps hajibeti learn how to respond. Imagine you are teaching someone how to write a poem. Instead of explaining every detail, you might show them a poem you would like to say. Make something like this with hijipiti. Showing examples or previous work or style help you understand what you are aiming for, whether that's writing in formal tone or being playful. Now let's talk about the output prompts. Have you ever followed a recipe that not only told you ingredients, but also showed you the exact way the dish should look? The format works the same way with AI. It helps Tajipti know how to structure it response. Imagine you are asking habit to help you create shopping list. Instead of just writing out ingredients, you could ask you to format the list into category like fruits, vegetables, and snacks. This gives the output a clear structure, making it more useful to you. Finally, we have trigger. This is like having someone a head start. You know how puzzle is easier when you already have the corner in place. A trigger is like nurse that points the AI in right direction from the get go. If you're writing a thank you note, you could start with a sentence. Thank you so much for your generosity and let a jib continue from there. It's like heading over the baton in Rs. Once you start in the right direction, it will keep going. Now, let's quickly recap the five components of the prompt. Instruction, information, example, output format, and trigger. These five pieces work together like a team, helping hat Jept understand what you want and how to deliver it. By using these components, you can turn a confusing or wag response to one that's clear and helpful. Think about it like having a great assistant who knows exactly how to follow her lead. Now, what I want from you is that now we learn what are the main five components which will give the better result. Take any problem you're facing right now and try to use this format, try to add these five components in your prompt. And after that compare how you get the output of general prompt, how you get output after applying these five components in your prompt. I also added examples so you can copy them, check them out, experiment with them, and also add assignment and do that and some to project section. I hope you understand what we are trying to do here and why these five compelers matter so much while writing prompt to get better results. So that's a portfoli will do, and I will see you in the next one. But before I go, I'm going to tell you something. So from now, we will understand the core concepts of AI, machine learning, recommendation or clustering. So we will see those AMAs topic in upcoming videos, and they will be mostly based on theories. And if there is any example, I will demonstrate them to you, and they are mostly theory based. So this is p, and I'll see where the next one is out. 17. Demystifying Machine Learning Concepts in ChatGPT: Hi, Ron Chetaner. Today we are talking about machine learning. So before we start, I want to tell you one's story from my college time. So when I was in college, I had this genius plan. Like, picture this. I was sitting in my hostel surrounded by mountains of textbook and past exam papers. I had what I thought was the most brilliant idea ever. You know, how we all try to guess what question will come in exam? Well, young me decided to full te wizard. Thought, why study everything when I can just create machine learning algorithm to predict the exam question. Classic or confident engineering student movement, right? So there I was dreaming about creating a magical algorithm that would crunch through years of question paper and boom, predict exactly what would show up on my exam. Talk about working harder, not smarter. But here's the money part. As I dove deeper into coding this masterpiece, reality hit me like turn of brakes. The exam was getting closer, my cold was getting messier, and I was spending more time deepugging it than actually studying. Looking back now, I can't help but laugh and think, Man, if only we had ha JP back in time, imagine dumping those all papers in ha GPI and saying, Hey, buddy, what's your best guess for the next exam? No Python coding required, no late night debugging session, no questioning my life choices at 3:00 A.M. But you know what? The whole experience taught me something valuable. Something that simplest solution like actual studying is the best solution. Although I have got to admit having air tool like ha Jep nowadays make our lives so much easier. Maybe just maybe don't use it to predicure question papers. You know, I studied hard for those exam, and I got outstanding mark or grade in that exam. Now, let's understand how machine learning works. You know, because of this machine learning, we are able to use models like Chat GPT. Have you ever wondered if you could teach a computer to learn without needing to be tech genius? Well, today we are going to show you exactly how you can do this using chat GPT. And by the end of this video, you will see how machine learning can be for everyone. Yes, even you. I remember when machine learning used to be this big mysterious thing. Some things only scientists in a lab could do. It seems so complicated, but now it's become so accessible that even my grandma could use it to help her find new baking recipe and she's 85. So what is machine learning? You may ask? It's simply teaching computers to learn from examples, kind of like how we teach kids to ride a bike. They practice, learn from their mistakes, and get better. The best part is, with tools like hagiPd, you don't need to code or build any fancy models. You could just provide examples and the air learns from them. So in Tolgvdo, we are going to explore how easy and fun it is to use machine learning within ha gibt. We will cover three main points like how machine learning has evolved and why everyone can use it now. Second point is how hagibt can help you get things done with simple prompts and examples. And third one is how you can teach the AI new tricks using examples from your everyday life. So are you ready? Let's dive in. In the past, teaching a computer to do something meant, gathering tons of data, cleaning it up and then training a very specific model, that could easily break if things weren't just right. You had to hire expert, and only big companies could really afford it. Imagine if you wanted to train a model to help sort your laundry. In the old days, you had need to collect lots of image of clothing, writing a program to recognize color and fabric, and spend hours of refining it until the computer got it right. But things have changed. Now we can use air like Chi DPD which is already ten on lots of information. Instead of starting from scratch, we just need to guide it using simple examples. No programming required, simple everyday tasks that we all do. And here's where things get exciting. Ja Gibt can perform tasks for you using just a prompt. For example, if you want to help you plan a birthday party, all you have to do is give it a few details like the theme, the age group, and maybe the kind of food you want. It will generate ideas, and you don't have to train a custom models to do this. It's all right there in the chat. Let's say you want you to come up with a few ideas for fun quiz game. You don't need to program or search through hundreds of website. Just ask Ja JBT. Can you give me three fun questions for three ware game? It uses the knowledge it already has to create something new for you on the spot. And here's the most exciting part. Not only you can use ha jib to get things done, but you can also teach it new tasks using examples. Let's say you want to train hagibd to help you write short poem for getting cards. You don't need to spend money or time building new air system. You just give it a few examples like here's a birthday poem and here's a thank you note. And it learns from that, or imagine you are shape and you want hagibt suggest recipe based on ingredients you have at home. Simply provide a few examples, stop your few dishes, and it can start recommending new ones that fit your style. So let's try one thing. Now you know how machine learning works and how you can use in ha Gibti. Now, there is one thing I was experimenting in ha JBT app. So, you know, like in Ja JBT, we can talk to ha Gibt, and it really talks like human. So I was experimenting, and I was asking some question. And it was, like, really human like. It was not like Alexa or Google Assistant. It was talking to me like human like. And I also tried a few things. So let me show you. Well, HIG PID, can you tell me how to make cake? And I only have two, three, four ingredients. Like I have biscuits, I have cottage cheese, as well as chocolate and milk. So can you tell me how to make cake using those ingredients? And tell me in such a way like you are breathing heavily? Okay. Let's see this. First, crush those biscuits into fine crumbs. Mix them with a bit of milk to form a thick paste. Now, spread this mixture in a dish to make the base. Melt the chocolate and mix it with the cottage cheese, still smooth. Pour this over the biscuit paste, then chill on the fridge sll set. And there you have it. Simple, no baked cake. So as you can see how Chad jupit is acting, it is breathing heavily and also giving us out. So this is really amazing. Let's let's try another example, what we ask. HR Jeopardy, can you write me bedtime story for like 10-year-old toddler? And when you are saying that story to me or when you are telling that story to me, talk in such a way like you are 5-year-old. Once upon a time. Hm. Okay, what happened next? In a big forest, there was headlines won't let me talk about that. Was Oh, I think I broke the code. Okay, but as you can see, it is working. And it all happened because of machine learning. But. So what we have learned today? Mochine learning is no longer for just expert. It's for everyone. Whether you want to help planning a party, writing a poem, or getting recipization, had Jeopardy can learn from examples you provide. The power of AI is now in your hands, and the best part is, you don't need to be a scientist to use it. With had Jeopardy, you can tap into motion learning and make it work for you in everyday life. I want to give you one task, open hat GPT app and try this thing, the new voice assistant thing. Is over here. There is a voice icon, and you can tap it. It will ask you, which voice it should be setting. Click on that and ask anything you want. And even though I also trying, it can speak in any language. Like other day, I said to Chat Jeopardy. Talking Mat and it was talking Mat very fluently. So that's a photo video. I hope you understand what we are trying to do here and what is machine learning and how you can use it without even writing single line of code. So that's a photos video. And in upcoming videos, we are going to see more advanced topics like lustering recommendation, and some other topics. So, so straight for that. I hope you are excited to learn those topics, and mostly they are going to be based on theory. There will be no, like, examples. They are mostly theoretic. So let's see what Z and see you guys in the next one. It's out. 18. Classifying Ideas and Data With Simple Prompts in ChatGPT: Did you know that you can perform tasks like classification? Something that used to required specialized knowledge and coding skill, which just promises true. Today, I'm going to show you how generative EI made this possible. Few years ago when I was in college, that time I was learning about AI. I was amazed at how complex it all seemed. You needed to be data scientist with the deep understanding of coding and algorithms to even get started. First four oh two today. And now anyone, even you can perform powerful tasks like classification. Always just a little bit of text, no coding required. So what is classification? Well, is the process of sorting things into category. A bit like organizing books on shelf. You could group them by genres like fiction, non fiction, fantasy, or by any label you want. And the amazing thing is, we can now do this with the AI and few simple prompts. In this video, I will walk you through how to use prompts to classify data in Tagepy whether you are dealing with survey, customer feedback or even organizing shopping list. This is going to blow your mind. O Let's start by understanding what classification mean in the world of AI. Think about a basket full of toys. You have got cars, dolls, and blocks all jumbled together. If you were to sort these toys, you might create plies for each type, one for car, one for doll, and one for block. That's classification. It's simply about assigning things to categories. Now imagine you have got a list of comments from school Feedback form. You want to group these commands into category like positive, negative or neutral. All you need to do is prompt Cha Jib to classify each command into one of this group, and guess what? You are using machine learning without even realizing it. Okay, so for this, I wrote this prompt. Like, please classify the following school felback comments into positive, negative, and neutral. And there are also sample student felback comments. There are 15 of them, and let's give this to Cha JP and let's see how Cha GPi classify it. So as you can see, Cha GP classified everything. But as you can see, there is a problem like in positive, it organized very well, but in negative, it is organizing, like, really weirdly. I think I have to write prompt again. Okay, but this is okay. The output is great. Now, let's ask Cha JBT. If you want to organize your data, if you want to organize the classified data into some format, like in column or table in a chart, you can also do that over here. For this, I wrote this prompt. Please organize these feedback commands into table and following with the following columns like ID, feedback command, and classification. So in this case, we don't have ID, but I think it will take as a number as a ID. So let's see I guessed it correctly, it is taking ID as a number, the number of this list, like one, two, three, something. And after that feedback and classification is positive and negative neutral. Now imagine if you want this data into CSU file, you can also do that. You can just add convert these commands into CSU format with headers like feedback, ID, command, sentimentals, and conference level. Okay, if you don't know what is CS, CSW comma separated values. So as you can see, let me show you. First GPT wrote in output the command, then it wrote the how is like the sentiments of the command. It is positive. And the perflx level is high. If you are a data scientist or data analyst, then you might be thinking, why I study data scienit or data analyst for this long? Now, Cha JPET can do within just like writing proms and it can do in seconds. I know there are lots of things in data analyst and data science, but these are like basic things you can do in Cha JP. I know there in future, there will be updates, and in that you can also produce graphs. So I was also watching some videos in that they was providing some data, and they was also providing actual graphs in ChaGPt. So we can also do that in Cha JPTy. Um, I was there are two or three friends who work in data analysts. They are data analyst, and I was showing them this technique, and they was also amazed. But they said to me, their company like don't allow them to use ChangePT. But they also use on their laptop and if they need it, and they complete their work. Now, let's take second example. Let's say you run a small business and you receive tons of emails from customers. Some are asked about product availability. Some are complaints, and others are positive testimonials. Sorting them one by one can be exhausting, right? Well, here's where prom based classification comes in. Instead of manually reading each email, you could copy all of them into hagibt and ask him to classify each email as inquiry, complaints, or praise. The AI will quickly sort through the text and organize the emails for you. No more inbox or load just simple efficient classification that saves your time and effort. For this example, this is my prompt, and I also added some details, the feedback about the store, and they are like this. And actually, I asked this to create some random data to Char JBD, and it created for me. That's why there is no email in this, but Cha JBT will arrange all it soon. Let's see, and how it works and what kind of output we get. Okay, we got praise. Okay, kind of, we are kind of doing sentimental analysis in previous course or in previous section, we saw the sentimental analysis. And this is exactly the same thing. And as you can see, hat J pretty messed up again. I gave complaints like this, and I wanted complaints like this. I think there is something going on with Chi Get today. Okay. But output is amazing. Like how we did in previous prompt, we organized that into table format. We also can do that over here, we have to write like this, analyze this customer message with ID, message type, classification, key phrase, emotion detector, and action required. And Char Jeopardy is working. So as you can see, hi Jeopardy works really amazing over here. So let me read the first output. So in our prompt, we provided this one as a first input. So that's why it took as a one. And after that, the message, then classification, then in classification, it is phrase, and the key phrases are amazing and all that. And after that, it detected the emotion. Even though we didn't provide any kind of, like, emotion detection in this still managed to do that. And then also gave us the action required, like thank customer for positive feedback. This is also good. I remember, like, there was my friend who was learning or doing some project on Twitter. They was collecting some data from Twitter, and they was kind of running the machine algorithm. And he told me in this algorithm, they added some key phrases because I was so curious about that technology and how they are doing. So he told me we collected all the key phrases, and we wanted to sort out or we wanted to analyze how many people behave badly or how many people commit bad words on Twitter. That's how they was using their algorithm, or that's how they was training their algorithm, using key phrases. For them, it took years to do that, maybe months to do that. But as you can see, with habit, we just did it in seconds, which just writing these simple prompts. Okay, I was thinking, let's take a real world application. Let's say, there is a LinkedIn post or something on LinkedIn. And I found this post like Google just drop J minimum 0.4 for the latest model and some information over here. And there are people who commented on this. So I was thinking, let's copy their comment, and let's do classification on them. And let's see what kind of are they positive about this thing or negative or the neutral. So let's do that. Let me copy all the data from here. So as you can see, I'm copying everything. So there is a name, the third plus and the four H 4 hours like reply, I'm copying everything, the proper picture. So in this, I'm not sorting anything, so let's copy this. But for this, I'm using Claude. Why I'm using Claude Because let me show you why. Now, if I page this file, so you can see Claude create a separate file. And from that, it's page the information and it runs the prompt. And I already wrote the prompt. Please help me classify the comments into response to my post. I want to classify the response into category, agree, disagree, neutral or others. And I also wrote another prompt in same prom. L produce the output into CSV into following format like job title, PY word summary and response to category. So let's give this to Claude. Nowadays, I use multiple LLM models like Cloud, hat Jeopardy, perpexlT and Gemini. So if there is any small dox I Gemini. If I want to do huge do task, then I use Cloud and hat Jeopardy. And if I want to search authentic information, then I use PerpexLT. So it is like I'm just jumping back and forth in LLM models for my purpose. Okay, so as you can see, we got the output, and so that's what I like about Cloud. So as you can see, it took data in this format and then it did its task, and then it gave us the output in separate section. Okay, so it created scopy like this Jota deal interns, then data scientist data is fun. Okay data, sop engineers and all that. And after that, five words, it means five words summary. Um, Okay. Okay, this is good. Like this is so this is great output. You can do the exact same thing in Chi GPT. But if I copy, let me show you if I copy that same data into this, so Cha JBT will do like this. And this is messed up, and if some person that person will get confused, that's by using Cloud. If you want you can use Cloud or Cha Jib, I totally depends on you. Both works like really accurate. So what did we learn today? Classification is an incredibly useful AI task that's now available to everyone thanks to prompt best tools like ha GPD and Claude. So as you can see, you don't need to be a data scientist and you don't need to spend hours of cleaning data and learning complicated algorithms. Just copy paste and let the AI do its heavy lifting for you. Whether you are sorting through feedbacks, emails, or product reviews, classification helps you organize data quickly and efficiently. And the best part is, it's all done with simple prompts. I know you want to try it yourself. Give it to go in hat Jeopardy and let us know how it works for you in the comments below. I hope you got the idea like how classification work in hat Jeopardy. Actually, we kind of doing machine learning without knowing it, and we are using words. That means prompts. So that's there's Porto Video. I added examples, so try them. A added assignment, do that and submit a preest. That's a Potvdo and I'll as in next one. Is up. 19. Take a break 3 Final video: D. Hi, everyone. I was making coffee for me. Sorry for the late. So I made filter coffee for me. It is quite famous in India, and it tastes so delicious. And by the way, you have completed 50% of this course. So take a moment to let that sink in and feel proud of yourself. And I'm also proud of you. You have made it further than so many people who started the course. Plenty of student buy a course, watch a couple of videos, and then life happens. But you you are here, putting in the work, investing your growth, and pushing yourself to learn this new skill. That's amazing. Seriously, give yourself a pat on the back because you deserve it. Now, if you have been powering through this course in one long setting or you've been glued to screen for hours, I have huge respect for that commitment. But right now, I'm going to suggest something important. Take a break. Here's why break are so valuable, especially when learning new skill when you take a step back, it gives your brain a chance to absorb all this information, sort it out and make those connections. This actually makes you better at learning and remembering what you are studying. Plus, break helps reduce stress and improve your focus. When you come back to tackle more content. So do your self affair, take a few deep breaths, stay away from the screen, maybe even take a nap. Go to for short walk. Or you can make simple coffee for you. This also helps. Go outside, soak up some sunlight or just do something that refreshes you. And, hey, if you want to share this world break with the world, feel free to post it on social media. And you can also tag me so I can see you are taking care of yourself while learning. Remember, you are not just here to rush through the content. You are here to truly learn and taking breaks is part of that journey. You are building skills that will impact your career and life. So no rush, just steady and strong progresses. So go take a break, recharge. And when you're ready, I will see you back here for the next part of the course. So this is photo Judo, and please take a break and go outside, meet your friends, and that's it. That's it. That's a photo Judah and I will see you again in the next one. Hey, sir. 20. Grouping and Clustering Content Easily in ChatGPT: Imagine you are a big family gathering with your cousins, uncles, and aunt. Like, how would you group them by their age, maybe by their hobbies or what kind of food they like. Today, we are dying into how Chidi Buti can do something similar, grouping data into clusters, which sounds tricky, but with simple prompts, you can do that in Jadi Bot. Remember the first time I came across clustering. It was like putting together a puzzle. At first, I wasn't sure which piece fits where. But once I found a common thread, everything started to make sense. In today's video, we are going to show you how to use prompts in Jag pity to do just that finding patterns and putting pieces together. Clustering is a way to group things that are similar to each other, just like how you might group family members by their hobbies. But instead of people, it can be used to organize ideas, comments, or even customer feedback in a way that helps us understand it better. So in clustering, we will cover three exciting things today. How clustering help us group things by sharing traits. Second one is a cool way to analyze survey response using clustering. And third one is how we can use prompts to automatically create useful groups, even for things like hobbies or job roles. So stick around. This will change the way how we think about organizing information in hatpit. So let's start with something simple. You know, when I was a kid, that time I used to do this like there was a box of crayons. So I used to cluster them by using the image on the box, and also sometime cluster them using the length. So this is what clustering is all about. Grouping item with something in common. Let's say you have box of buttons, you could ask AGBT to cluster them by colors. It might group red buttons, blue buttons, and green button into separate cluster. This kind of simple clustering help us to see pattern we might otherwise miss. Now let's go deeper. Imagine you had just ask your classmate how they use their free time. Some might say they love reading, others prefer playing sport or some enjoy painting. Instead of sorting them one by one, you could ask hagibdy to cluster the answer. It might create groups like creative activities, physical activities, or intellectual activities. So I have some data as well as the prompt for this clustering. So I wrote this kind of clustering prompt for Ja GPI. This the following response about hobbies into groups like allow reading. I enjoy playing sport, painting, and I'm using puzzle or basket boss board. And also added data of of students like this. There are like ten or 15 students. Data I to from JAGPI these are random data. If you actually want to take data, you can use Kaggle. In that website, um, when I was learning data science and machine learning when I was in fourth year like two or one year ago in college. So that time we used to take data to train our models or maybe practice machine learning or AS skills. So if you want to, um, do that, you can go to Cagle. I will add the link in the resources, go there, and take some data and add it over here and do some clustering classification or machine learning things on it. So let's give this data to ha db Okay, so ha GPT for sports, then creative activities in this sports where Alex and Mia, they like playing soccer and basketball. We can also cluster it further, like let's say in sport. We can specify if the student plays only this kind of sport like football, then only cluster this kind of student into football. Also if that person or student play basketball, then cluster them into separate section. You can also do that as well. So this is good. I hope you got the point how clus stream works and what we are trying to do here. It's kind of classification, but in this, we are grouping same interests in one group. Now let's move on something really powerful. Using clustering to understand last set of data, like customer feedback or job roles. Imagine you are running a business. You want to group customers by what they like most about your products. Cha JB can automatically find common themes or traits in their feedbacks. So for this example, I took toy company, and I took the feedback from customers like what they are seeing about their toys and this kind of survey. And for this, I wrote, analyze analyze and cluster the following customer feedback into categories. Like, these toys are really durable. My kids learn a lot while playing this toy. Toys or toys are so much fun. So these are some customer feedbacks. So let's give this to hit Deputy and let's see. First one is durability and Sept. Okay, the toys are really durable with. Okay, this is good. It also has educational value. Some customers said it has educational value. And for engagement, imaginative play. Okay, this one is good. And so as you can see, had by also cluster this one as well. Okay, so in previous udio we took some data from LinkedIn about one post on one AI post about Gemini 1.5, launch of Gemini 1.5. In that we added classification. So I was thinking or I'm currently thinking that let's add clustering into this. And for this, I wrote this prompt like create testimonial of the job roles of people, and that's that response. Like output dime to ask a tree. So I told you you can format data into any format, you name it, you can format into hat GPT. Let's see how hat GPT gives output for this format for this clustering. Okay, so we got our ASK tree, and First on technology are also people who work in software development, like Soft Engineer and software maintenance. After that AI and data, then we have data scientist, data engineer, AI, all that. Okay. Okay, we also have education and training intern. That's good. And we also named the companies OE. Self described. So as you can see, clustering is really amazing. So I was saying in previous example like this toy example or maybe in this skills example, the activities, what do what kind of activities they do after school. So I was saying that we can, clusterize I was saying if they like sports, in that if they like basketball, then again, we will cluster them. But using this aske tree can do the same thing. We don't have to write a prompt again. So that's the benefit of using the asked tree. Or different kind of format. You can also use CSU table, you name it, you can illster it in any format. So today, we have learned how clustering help us group things that are similar from simple objects like and hobbies to more complex ideas like customer feedbacks. By using habit and the right prompts, we can actually disco patterns in data that would be hard to spot otherwise. Clustering help us make sense of the world around us, from organizing survey response to figuring out what our customers care about. And the best part is we don't need to be a tech expert to do this. With habit, anyone can start clustering with just a few simple prompts. If you want to learn more about how to apply clustering in different ways, what's one way you could use clustering in your life? Could you group of your books or maybe organize your household items? Share your idea in commands. So for this, I added some examples, copy the page them, experiment them, do that. After that, also a SN, do that seminar to section. I hope you got the idea, what we are trying to do here. And that's it that's port video and do this apply clustering, classification, um, take any real life example and um add these prompts in that, clustering prompts, and cluster data, and apply different formats to it, like we applied Ask t. You can also press B. You can add any kind of format. So that's a photo judo and I'll see then the next one. This out. 21. Making Predictions Based on Prompts in ChatGPT: Hi, Rwandanir. You know, when we was learning about machine learning in first video of this section, like I told you the story about how I was building machine learning algorithm for my quien papers. I wanted to predict future questi papers by using previous question paper by feeding that data. So today we are doing kind of same. We are using prediction in hagiputting. What we are doing with the help of proms, not using coding. So today we're down into a world where your question becomes crystal balls. Imagine asking your friends about tomorrow's weather and they actually get it right. That's the power of prediction which hagibity prompts. And, trust me, it is going to blow your mind. You know, I remember when I discovered this in Cha ibt, I was trying to figure out what movie to watch on Friday night. And instead of schooling endlessly on IMDB, on Google, I asked had gibt. The suggestion was spot on, and I had the best movie night ever. So why is this important? Well, in a world full of choices and information, having a tool that can help us make smarter guesses about the future is like having superpower. And the best part is you don't need to be a tech wizard to use it. Today, we will explore this amazing way you can use Cha Gibt for prediction. So get ready to see the future in whole new light. So I love to read a book like novel, self help, fiction, non fiction, as well as biography. So one day I was, like, reading Harry Potter. I know I was very late at reading that novel. And like after reading that whole series, I wanted to read something else in this fantasy world. So I asked Agibet. Like, I asked, like exactly this prompt. Like, based on my love for Harry Potter, three books I might enjoy because it's the magical element coming of age story and friendship theme. I know there isn't not that amount of examples over here. I just put Harry Potter. Still had Gibeti gonna predict a book like your next book. Okay, so Chad JBT gave us three suggestions post Perky Jackson and the Alum peers by Rick Riordunt. Actually, I saw this movie. It used to be on Disney, when I was, like, in school time. That I used to watch this movie also was I guess his dark materials just came on the OTday. It is on HBO or something like that. And Miss Per Green's home Opera. Okay, this is I didn't read it. I will search for this book and I will ask J Revis and all that. Then I will learn how to read from Amazon, and then I read this book. See how it uses what you like to predict, what you might enjoy next. It's like having a super smart librarian in your pocket. Okay, now let's take a second example. In that, I was thinking, let's take some real world application. Let's say, let's say you are you want to work in IT industry or in AI field or maybe in renewable energy. So you can ask IGBT. Based on current technology trends like AI and renewable energy, predict five jobs five jobs, skills that will be in high demand in next ten years. So let's give this to RGB. Okay, so first one is A and machine learning. After that, we have data literacy analysts. Okay, these are really amazing field to work with. And I have few friends who work in this field. I also used to work in this field, and now I teach prompt engining. Still, the work I do, the prompt engineering and all that, it is totally related to these two fields. We also have renewable energy after that. Cybersecurity is also great. But in cybersecurity, I saw people, like, told me, uh, I don't know about region, but in India, they said they don't pay you that much. So before, if you want to go into any career, do lots of research it and then jump into that career. And last one is adaptability and cross function collaboration. Okay? I don't know anything about this. Am Okay. Okay, this one is also great. If you want to read, pause and read this one is also great. Now, let's go to Cloud and let's try that our LinkedIn example. So I want to ask Cloud. Like we already fed lots of information in Cloud, and we have several prompts to it. Like, first, we like classify data, then we also organize data with the help of asketr. Now, what I want to do is that I want to predict which industry are not available over here. See most of industries are over here, like, creative field also their businesses also their educational also there, as well as technology also there. But I'm not seeing the medical field. So let's ask loud and let me see what kind of industry is not in air right now. According to our LinkedIn data, I know I'm not taking that a huge amount of data, but still, let's see what kind of um industries there are, which are in not you know, like I research a lot. I make lots of videos on AI on YouTube as well as on Instagram. So I saw the lots of AI work is happening in, like, medical industry. So I was telling in my previous course. So there was one AI. They trained some pictures about people who added breast cancer. And that algorithm detected breast cancer. Even doctors was not able to, like, detect the breast cancer, but still that algorithm detected the gas in that person. So let's read the output. Firstly, research and academia, second is healthcare, third is legal service, okay, fourth financial service, and we have content ama production. Okay. But I have to say in content and media production, people are using AI like crazy. So there was writers strikes happening like a few months ago. Writer was so worried about like this AI integration in their industry. They started protesting in Hollywood. But you know, why it is saying continual media production because so in this LinkedIn data, lots of people from that field didn't comment it. So that's why it is predicting that. So for this algorithm to work or for this prom to work, we have to provide a bunch of data. Then it will be then it will give us the accurate reason. We can't blindly follow the prediction. Still, the result is accurate based on this data. So as you can see how amazingly it works in Cloud or Mybin as well as in Chat JPITy. So like a few days ago, I was experimenting with this Cloud, as well as with Chat JPET. So I put all the name of my courses, and I was confused, like what next course should be I working on. And so I provided data, I also provided my previous course name, as well as some information. Like, there was like, like ten pages of information, and then I asked Lau currently, I have courses in this field. So I was thinking to go in this field, and I also this amount of knowledge. So can you suggest me what course I should I be working on? He suggested me like you should teach meditation and all that. Why Clause saved me that? Because I provided the data like I also do meditation. I have, three, four years of experience in it, also have intermediate fasting and all that in a workout. So that's why he recommended me three or four options like meditation, teach workout, as well as intermediate fasting and how to plan diet and something like that. I was going totally in opposite direction because I mostly created like 100% 100% of time. I created courses on technology field but I'm also interested in that field. So that's why Claude recommended me to make courses on that. And the prediction was really accurate because from a few days or few months, I was thinking I should be making courses on those fields as well. So it predicted really accurately. But to get accurate prediction, you have to provide lots of examples, lots of data. Then you'll like predict accurately. So in my example, I provided like ten pages of data, the personal data, that. So that's why it was able to predict that information know, in this video, we just took like a basic example we asked what book should be I'm reading next and also about movies, as well as we also predicted this linked in comment section. So we are doing really simple tasks. These are not really huge tasks. If you actually go for actual machine learning code, if you saw those, you mind will blow away because those models are totally different. And, you know, like I was talking with my friend, and I asked her, like, why you don't use Chi GPT in your company? So that person said, We don't want to breach our data with HIGPID. Char and what ChIGPt Cloud does is that it learns from the data which we provide. So that's why they don't want to bridge the information about that company. So if they have small tasks like this, they can use this but they have huge amounts of data and all that, so they can't use the Cloud and Chi GPT. And there you have it, folks. We have journey from predicting your next book and even peeking into job market of the future. All of this with just a few well crafted proms to charge pity. Remember, the key to good prediction is asking the right question. Think about what you already know and use that as straight point for your promise. So what will you predict next? Maybe you will forecast the new big vacation spot or the next viral dance craze. The possibilities are endless, and the future is yours to explore. Oks added some examples, try them out, compute them, try them out, and do experiment on them, add things you want to, like, experiment with the layer, and try to experiment with them, like, add your own data and try to predict something using Cloud or chat GPT. And I also added assignment, so do that. I know in this we are not totally focusing on the prompts, like how we should be writing prompts. But the main thing I want to teach you that how chats better work, how it analyze, how it responses. So that was the main thing. And I don't think so we need to write the huge amount of prompts. But if there is, if you want to really want to extract some amount of data or really want to predict something, you have to give some examples. I'm just scratching the surface over here. So that's why I'm taking small, small examples, and I just want to teach you how this works. So that's why I'm just taking small scale example. That's it. So that's a photo video. Until next time, keep predicting, keep exploring, keep experimenting and keep writing prompts. So that's a port Video, and I'll see you and then the next one up. 22. Personalizing Recommendations Using Prompts in ChatGPT: I want to ask you one question. Like, have you wondered how your favorite online store seem to know exactly what you want to what you want to buy from their store. And like you also noticed on Netflix, like it also recommend you like one of the best movies, as well as on social media platform. After watching real or Tik Tok, it recommends the video related to that or maybe or according to your choices, maybe on you to also recommend if you're watching lots of content on AI, then next video will be on AI. Or if you are reading watching book summaries, then it will recommend you the channels to make videos about book summaries. I know you had noticed a lot. If you do, please let me know in comments and I also want to recommend you a movie called Social Dilemma. I know lots of people must have watched it. It is also available on Netflix. I think it is one of the great movie. I'm recommending it because how this social media company, like, hooked you on their platform for a long time and how they recommend you sturb. So that's why I'm recommending that movie. So Go ahead and watch that movie, and please let me know you can DME or social media platform you like that movie or not. So in this video, we are going to learn about recommendation. I remember when I first learned about this in 2016 when I was in college, I felt like I had discovered the secret superpowers. And today, I'm going to share that superpower with you. We are going to explore how Chat GPD can help us make smart solution, just like your favorite apps do. By the end of this video, you will see the world of recommendation in whole new light. So get ready to be amazed. Imagine you are at library and your friend loves a book about dinosaur. The librarian might say, Oh, if you like that, you might also enjoy this book about Ancient Egypt. That's exactly what we are talking about, suggesting something based on what someone like. Now let's see how Chi JEPIt can be our helpful librarian. Say we have a group of friends who love different ice cream flavors. So in this case, we can ask to Chi JEPIt Los at who likes chocolate ice cream, find out other people who didn't mention chocolate, but might enjoy it based on their similar taste or preference. I provided some data to it, and if a few people likes it, it will separate the data. And if a few people like the ice cream, then we can recommend the chocolate flavor ice cream. So from this data, Chad Jeopardy says, Sarah, Jams, Emma, Ol and Lima have explicitly profsted their own love for chocolate ice cream. So these people love chocolate ice cream. But, um, Chad Jepite says from the others, there are solid chances they are chocolate fan based on their flavor preference. Like Michael loves rich and creamy dessert with intense flavor. So in this case, we can recommend Michael chocolate flavored ice cream, as well as Sophia, as well as Lucas, as well as Isabella. So this is one of the basic example about recommendation in Cha Gibt using compte engineering. Before we jump on to the next example, I want to show you something. So this is Amazon account. And like from a few months, I'm searching for Smart for, and there is a festival going on in India. So it is recommending me a few devices. Like it always whenever I open the app, it always recommend me this Smart f. So DVA is in a few days, so it also recommend recommending me some lights. So for DVLI because D is at festival of lights. That's why it also recommending me the lights. Now, let's jump onto a second example. Here's where it gets really exciting. Chad Jeopardy can use Context to make even smarter recommendation. Let's say we are planning family vacation. So in this case, we can ask had Jeopardy, like, the Smith enjoyed their beach vacation in Florida, looking at all the families preference and budget, such a similar vacation they might like. I provided a bunch of recommendation or bunch of reviews from some websites. I took reviews from other websites, and the people who like traveling, they added these kind of reviews about their places. So they also added, affordability, as well as attraction, as well as it is family friendly or not. So we have this amount of data, and in this case, we are adding contexts. So Cha Je Put will work on that context, and it will provide us the accurate output or accurate recommendation for our next so let's give it to JIGPD. Okay, so hi GPD recommended us a lot of options like there are eight of options we can choose from. And we can also ask, like for them. I want only one location, and this is my budget, and this is my family this amount of people are coming with me. So then it will sort only one. You can ask AJPdGive me only one. I don't want eight of the, and it has to be close to my location. If you give this kind of prompt, it will recommend you one. But as you can see, by considering multiple factors, hi Jeopardy help us make thoughtful and personalized recommendation. And there you have friends. We have uncovered the magic behind recommendation from suggesting books and asking flowers or planning a vacation, Cha J PD can help us making smart personalized suggestion just by asking the right question. So let me tell you how I use the recommendation in my daily life. So, you know, I make courses. I have 11 courses on multiple platforms, and let's say, I want to promote my courses to, like, several people. So in this case, I give the data, like the platform gives me data. Like you have this amount of students from this region, and they do this kind of job. And, um, so I provide that amount of data to Cloud at Chit GBD, and I ask hA GPD. So I have this amount of data. Can you recommend or can you suggest me like, for whom I should be recommending my courses through mail? So ChaGBT and Cloud analyze that data and tells me you should recommend or you should promote courses to this specific industry or this kind of or specific to this age or maybe specific to their field. So this is how I use Cha GBT for recommendation and as well as to promote my courses. But remember, the key is to provide relevant information. Ask Ja Jib to find connection. It's like using a super smart friend who remembers everything and can spot patterns in a flash. So next time you are wondering what movie to watch or what gift you buy for your friend, try asking ha Jib for recommendation. You might be surprised at how helpful it can be. In last video, I hope I recommended you a movie called Wild Robot, and I hope you was that movie. If you're not, please watch it, because it is kind of related to prompt engineering. If you watch that movie from perspective Prompt Engineer, then you will know how robot learn and how things work in this AI. So that's why I'm recommending you, and that's why I'm forcing you to watch that movie. I hope you learn how recommendation work and how we can use in hat GPT. What are some application. And let me know how you are going to use this recommendation, like in your daily life. So that's what video, and I will see you guys in the next one. Is up. O. 23. Teaching Models Through In-Context Learning in ChatGPT: Imagine being able to teach an AI, new task without writing single line of code or fitting terms of data. Sounds like magic, right? Well, this is possible through concept called in context learning. Today, I'm going to show you how you can train AI models like Chat GPD or Claude using just example, making complex task a breeze. When I first started working with AI, I thought training model would involve lots of math, coding, and advanced algorithm. And that was true. So when I was in college, I used to do these things, like I used to do math, coding, as well as I used to apply algorithms. But one day I stumbled upon a method. Like, I was just experimenting with hat Jeopardy. So I was like, I college time, I used to do this. So, can I just use prompts to do the same thing in chat Jeopardy? So that time I stumbled upon that felt more like showing someone a few examples and watching them master the task. It completely changed the way I saw a training. This method is called in context learning. It's like giving Cha Jeb a quick lesson on a topic and seeing it picks up the pattern immediately. And the best part is, it requires no complex programming. Today, we are going to break it down into bit size pieces so anyone can understand it. By the end of this video, you will learn how to use in context learning to refine your prompts, create better outputs, and even perform advanced tasks like using examples. And we will be doing it with some surprising use cases. Trust me, you will be amazed at what you can achieve. So let's start with a straightforward example. Imagine you have a list of random grocery items. Some of them have long and detailed names like freshly harvested organic baby carrots or extra virgin olive oil imported from Italy. Now, let's say you want to train hagibei to simplify this description to just carrot or olive oil. This is where in context learning comes into play. Have to show Cha Jipidi a few examples of what I want. Like in input, I can type like frislyharvesed organic baby carrots and input, I will write carats. And the second example will be input extra virgin olive oil, imported from Italy, and output will be olive oil. Now, when I ask ha JepIdi to look at whole grain bread with sesame seed, it simplifies into bread without any additional instruction. That's the magic of in context learning. Okay, so I added some data into Cloud. You can use ha GPD as well. But I use Cloud, like it gives the data output in better format. That's why I choose Cloud. Ja Jibit also work exactly the same. So it totally depends on you. If you want to use ha Jept if ha JBT or Cloud, maybe Gemini or maybe per pixlty, it totally depends on you. So I asked Tha Jeopardy. Your task is to simplify detail organized product name to their basic form. And here are some examples. So I added some examples like handpick premium, Washington State trade Delicious Apple, and then I just wrote Apple. If normal person, like, read this the hand pick premium, Washington, red, less apple, that person is, like, What should I be buying? But, if that person is shape, so that person will know, like, we have to buy apples. But I know the breed is totally different of the apples, but we can ask to share our shopkeeper. Um, Okay, after that, I also add a few examples. You can pause, can read it. And let's give this data to Claude. Claude will learn from that data. Okay, I hope Claude is learned from that data, and it also saying, should I remove descriptive adjective like this? It is saying it is suggesting some tips for us, but we don't want that. And it is also saying, like, what is my goal? Like, would you like me to simplify some more product names? So it is kind of confused because we didn't mention, what should a Claude be doing with this data. But now I hope it learns from data, and it always learn from that data. So now I said, now simplify the following product name. We give some data. So let's see, let's give it to Claude. So as you can see, the output is really how I want it. But we can modify this like in first example, there is a freshly squeezed hue orange juice with extra par. So in this case, it should be writing orange juice, not just juice, but it's still rot juice. It is okay, because we didn't provide that amount of information to it. But still it learned on its own. We didn't mention in detail, like, as you can see, I just wrote now simplify the following products. Still it work like really amazingly. Now that we have seen how it works with simple text, let's level up a bit. Imagine you are working with a table of movie rating. The table has column for movies names, genre and ratings, and you want hagibd or Cloud to predict the genre of movie based only on the rating and the number of reviews it has. Normally, you need a whole statistic model to figure this out. But with incontext learning, you can just show hagibt rows of data. Okay, in this front, I added my all the favorite movies like Dark Night, Inception, ShanshakRdemption, the hangover, all my favorite movie added in this list. If you want to watch this movies, you can pause the video and get some recommendation from me. These are like cult classic movies. And at the end, I also added some movies from that Claude will predict. Like what kind of what is the genre of that movies? So I did everything in one prompt. I don't have to read a prompt again and again. Like in the first case, what we did is that. First, we gave the prompt as well as the data, and then it got confused, like what should I be doing with this data. Then I told Claude, you have to do this with this data. So Claude was confused for a while, but in this case, I'm not doing that. So I'm giving all the data into one prompt. And I hope it works. Okay, so Claude really predicted the genre, very accurately. So as you can see, I think we all the interest teller, and it belonged to Sci Fi. And after that, we are godfather. That is drama. And it also mentioned, like, why it belonged to Sci Fi. And so as you can see, highest rating for fd is consistent with other sci fi fields like inception and metric. Second point is, do you and count around 1.9 million matches scale of the other Sci Fi production. So from that, it peaked it could be Sci Fi. And after that, it did the same with Godfather, as well as with the Super Bad and as well as with the oil place. Actually, I saw all of this movie, and I have to say the genres are really accurate. As you can see, we did this without writing single line of code without writing a single line of code. We just gave simple data to it, and from that, it predicted the genres of that movie. So as you can see how amazing these LLM models nowadays are, from this technique, if you like, own business or if you have a huge amount of data or if you want to do something with data you can literally do anything with data. You can format in different way. You can analyze that. You can, like, add recommendation onto it, or maybe you could say to it, I have data of my customers, as well as I have data of my patients. They eat lots of junk food, and I also recorded their behave. So if next person comes to, if you just provided some data about that person, it will tell why that person eat junk food. Actually I did that experiment. I took data set from Cloud, and I took data from Cagle and past it into Cloud, and I asked same thing. I asked same thing to Claude and the details was really accurate. I remember when I was in college, I used to do this kind of stub, like machine learning. That time we was with my friends, we used to analyze the data we have. So that I remember, it used to take us like four or five days to do simple task, and the code was also huge approx hundred to 500 lines of code. I totally It depends on the project, but on average, it was like that. So as you can see, in this case, it took us just like one or 2 minutes. And we just provided really small amount of data. Our case, we actually gathered lots of data from Kale and other websites as well, and then we wrote algorithm, then it predicted. So that's how in context learning work, and that's how we can train LLM models like Claude and Char Jedi to learn from new data. So to wrap it up in context learning allows us to teach charge Body or cloud new tasks using just a handful of examples from simplifying long description, preting genres on rating or to categorize product on e commerce platform, the possibilities are endless. This method can empower anyone, even those without a technical background to leverage AI in ways that never thought possible. Whether it's automating tidyless tasks or generating insightful prediction in context learning opens up a new opportunity. You found this video helpful, please let me know. And let me also know how you are going to use this technique to predict data. Going resources, add a few examples like how you can ask same task in a different way. So as you can see on screen, we took the example of movies. I also ask in different ways like this, as well. I wrote three kind of examples like type three ways you can ask to cloud or hagibty. So copy them, paste in hagibt or Cloud and try to experiment with them, also add your own data, and try to teach Chagbt or cloud from that data and predict some output from it. Also add sein, so do that seminar project section. Remember, AI is not just for tech expert. With tools like Cha Jeopardy and Cloud, anyone can become prompt engineer. So keep experimenting, keep learning, keep writing prompts. So that's it. That's it photos video, and I will see you guys in the next. Is out. 24. Performing Classification with Prompts: Did you know that you can perform tasks like classification, something that used to require specialized knowledge and coding skill, which just promises true. Today, I'm going to show you how generative EI made this possible. Few years ago when I was in college, that time I was learning about AI. I was amazed at how complex it all seemed. You needed to be data scientist with the deep understanding of coding and algorithms to even get started. First four oh two today. And now anyone, even you can perform powerful tasks like classification. Always just a little bit of text. No coding required. So what is classification? Well, is the process of sorting things into category. A bit like organizing books on shelf. You could group them by genres like fiction, non fiction, fantasy, or by any label you want. And the amazing thing is, we can now do this with the AI and few simple prompts. In this video, I will walk you through how to use prompts to classify data in ajepdy whether you are dealing with survey, customer feedback or even organizing shopping list. This is going to blow your mind. Let's start by understanding what classification mean in the world of AI. Think about a basket full of toys. You have got cars, dolls, and blocks all jumbled together. If you were to sort these toys, you might create plies for each type, one for car, one for doll, and one for block. That's classification. It's simply about assigning things to categories. Now imagine you have got a list of comments from school Fallback form. You want to group these commands into category like positive, negative or neutral. All you need to do is prompt Cha Ji to classify each command into one of this group, and guess what? You are using machine learning without even realizing it. Okay, so for this, I wrote this prompt. Like, please classify the following school felback comments into positive or negative, and neutral. And there are also sample student felback comments. There are 15 of them, and let's give this to Cha JP and let's see how Cha GPi classify it. So as you can see, Cha GP classified everything. But as you can see, there is a problem like in positive, it organized very well, but in negative, it is organizing, like, really weirdly. I think I have to write prom tug in. Okay, but this is okay. The output is great. Now, let's ask Cha JBT. If you want to organize your data, if you want to organize the classified data into some format, like in column or table in a chart, you can also do that over here. So for this, I wrote this prompt. Like, please organize these feedback commands into table and following with the following columns like ID, feedback command, and classification. So in this case, we don't have ID, but I think it will take as a number as a ID. So let's see I guessed it correctly, it is taking ID as a number, the number of this list, like one, two, three, something. And after that feedback, and classification is positive and negative neutral. Now imagine if you want this data into CSU file, you can also do that. You can just add convert these commands into CSU format with headers like feedback, ID, command, sentimentals, and conference level. Okay, if you don't know what is CS, CSW comma separated values. So as you can see, let me show you. First GPT wrote in output the command, then it wrote the how is like the sentiments of the command. It is positive. And the perflx level is high. If you are a data scientist or data analyst, then you might be thinking, why I study data scienist or data analyst for this long. Now Cha JPET can do within just like writing proms and it can do in seconds. I know there are lots of things in data analyst and data science, but these are like basic things you can do in Cha JP. I know there in future, there will be updates, and in that you can also produce graphs. So I was also watching some videos in that they was providing some data, and they was also providing actual graphs in Cha JPETy. So we can also do that in Cha JPTy. Um, I was there are two or three friends who work in data analysts. They are data analyst, and I was showing them this technique, and they was also amazed. But they said to me, their company like don't allow them to use ChangpT. But they also use on their laptop and if they need it, and they complete their work. Now, let's take second example. Let's say you run a small business and you receive tons of emails from customers. Some are asked about product availability. Some are complaints, and others are positive testimonials. Sorting them one by one can be exhausting, right? Well, here's where prom based classification comes in. Instead of manually reading each email, you could copy all of them into hagiby and ask him to classify each email as inquiry, complaints, or praise. The AI will quickly sort through the text and organize the emails for you. No more invoxO load just simple efficient classification that saves your time and effort. For this example, this is my prompt, and I also added some details, the feedback about the store, and they are like this. And actually, I asked this to create some random data to Char JBD, and it created for me. That's why there is no email in this, but Cha JBT will arrange all it soon. Let's see, and how it works and what kind of output we get. Okay, we got praise. Okay, kind of, we are kind of doing sentimental analysis in previous course or in previous section, we saw the sentimental analysis. And this is exactly the same thing. And as you can see, had Ji pretty messed up again. I gave complaints like this, and I wanted complaints like this. I think there is something going on with hi Get today. Okay. But output is amazing. Like how we did in previous prompt, we organized that into table format. We also can do that over here, we have to write like this, analyze this customer message with ID, message type, classification, key phrase, emotion detector, and action required. And Char Jeopardy is working. So as you can see, hi Jeopardy works really amazing over here. So let me read the first output. So in our prompt, we provided this one as a first input. So that's why it took as a one. And after that, the message, then classification, then in classification, it is praise, and the key phrases are amazing and all that. And after that, it detected the emotion. Even though we didn't provide any kind of, like, emotion detection in this still managed to do that. And then also gave us the action required, like thank customer for positive feedback. This is also good. I remember, like, there was my friend who was learning or doing some project on um, Twitter. They was collecting some data from Twitter, and they was kind of running the machine algorithm. And he told me in this algorithm, they added some key phrases because I was so curious about that technology and how they are doing. So he told me we collected all the key phrases, and we wanted to sort out or we wanted to analyze how many people behave badly or how many people commit bad words on Twitter. That's how they was using their algorithm, or that's how they was training their algorithm, using key phrases. For them, it took years to do that, maybe months to do that. But as you can see, with Chip, we just did it in seconds just writing these simple prompts. Okay, I was thinking, let's take a real world application. Let's say, there is a LinkedIn post or something on LinkedIn. And I found this post like Google just drop J minimum 0.4 for the latest model and some information over here. And there are people who commented on this. So I was thinking, let's copy their comment, and let's do classification on them. And let's see what kind of are they positive about this thing or negative or the neutral. So let's do that. Let me copy all the data from here. So as you can see, I'm copying everything. So there is a name, the third plus and the four H 4 hours like reply, I'm copying everything, the prople picture. So in this, I'm not sorting anything, so let's copy this. But for this, I'm using Claude. Why I'm using Claude Because let me show you why. Now, if I page this file, so I can see Claude create a separate file. And from that, it's page the information and it runs the promt I already wrote the prompt. Please help me classify the comments into response to my post. I want to classify the response into category, agree, disagree, neutral or others. And I also wrote another prompt in same prom. Like produce the output into CSV into following format like job title, Pi word summary and response to category. So let's give this to Claude. Nowadays, I use multiple LLM models like Cloud, hat Jeopardy, perpexlT and Gemini. So if there is any small dox I Gemini. If I want to do huge do task, then I use Cloud and hat Jeopardy. And if I want to search authentic information, then I use PerpexLt. So it is like I'm just jumping back and forth in LLM models for my purpose. Okay, so as you can see, we got the output, and so that's what I like about cloud. So as you can see, it took data in this format and then it did its task, and then it gave us the output in separate section. Okay, so it created sphal like this Jota deal interns, then data scientists data is fun, okay, data, sop engineers and all that. And after that, five words, it means five words summary. Um, Okay. Okay, this is good. Like this is so this is great output. You can do the exact same thing in Chi GPT. But if I copy, let me show you if I copy that same data into this, so Cha JBT will do like this. And this is messed up, and if some person that person will get confused, that's by using Cloud. If you want, you can use Cloud or Cha Jib, I totally depends on you. Both works like really accurate. So what did we learn today? Classification is an incredibly useful AI, task that's now available to everyone thanks to prompt best tools like Chagpd and Claude. So as you can see, you don't need to be a data scientist and you don't need to spend hours of cleaning data and learning complicated algorithms. Just copy paste and let the AI do its heavy lifting for you. Whether you are sorting through feedbacks, emails, or product reviews, classification helps you organize data quickly and efficiently. And the best part is it's all done with simple prompts. I know you want to try it yourself. Give it to go in hat Jeopardy and let us know how it works for you in the comments below. I hope you got the idea like how classification work in hat Jeopardy. Actually, we kind of doing machine learning without knowing it, and we are using words. That means prompts. So that's there's Porto Video. I added examples, so try them. A added assignment, do that and submit a presession. That's a PotoVD and I'll as in next one. Io. 25. Choosing the Right Examples: How Many and Which Ones in ChatGPT ?: Have you wondered how many examples are too many examples when using AR such GPT or Claude, or maybe you are unsure if a few examples are enough to get the AI to understand what you want. Well, today we are going to uncover exactly how to determine the number and type of example you need for an effective prompts. Okay, when I was learning machine learning in my college, so that time, there was one, really amazing saying was famous. Like more information you fail to model the accurate the result. But I remember when I first started exploring prompt engineering, I thought, why not just throw a bunch of examples at it and see what happens? Turns out that's not always the best approach. I ended up with massive confusing mess that made the AI more perplexed than I was. Since then, I have learned a new tricks that can save you from the same headaches. Alright, let's dive into the art of choosing example for your prompts. The goal is to keep things simple yet effective. So I will break it down into three easy to follow session. By the end of this video, you will know how to make your prompts a clear and produce more accurate outputs. First up, let's talk about types of task you might use Char GPT for and how many examples each one needs. For simple tasks like sorting item into category, let's say sorting pets into mammal, reptiles and birds, you really don't need a lot. Three or four, for example, like, dog is mammal, a snake is reptile, and parrot is bird are enough because this L&M model already knows and they already trained on this data. So we don't have to provide that amount of data to plod or charge a bit. If you give too many examples like listing 50 different animals, the model might get overwhelmed or start predicting every animal as something similar to what it so frequently. Like calling every animal a mammal, so keep it balanced. Now, what about your complex task? Imagine you are trying to get the model to understand nuances in language, like identifying different tones or emotion intakes. For this, a few examples might not be enough. You will need to provide various examples like showcasing wide range of tones like happy, sad, sarcastic, natural, and so on. Say you want to identify sarcasm. Instead of just saying John said nice job sarcastically, you could include several different scenarios like John glanced at the broken and said, Great work, genius, or Sara look at the empty fridge and said, wonderful planning. These verified example help the AI capture subtle differences. But there's a cache. The more example you add, the more computational resources you use. If you are using the free virgin or even premium plan, you might run into limitation of context length, and that's something to keep in mind. And here's the million dollar question. Is it better to have more example or just a few really good ones? I will say it all about quality. If your examples are clear, verified, and captured all necessary aspect or task, you won't need many. Imagine you are teaching Tajipt how to identify famous paintings. Instead of giving hundreds of examples, pick 1020 that span different styles like impressionism, surrealism and abstract, include details like the artist's name, the painting style, and one defining feature. This focus set will train the model better than a random collection of numbers. In one of my project, I only use ten examples for categorizing different writing style, and that turned out to be more effective than using hundreds of random sample. This saved me time, resources, and improve the accuracy significantly. To summit a four straightforward task, a handful of examples will useful to do the trick. For more complex tasks aim for the diversity and coverage in your example. Rather than sheer volume. The third one and most important one is quality over quantity is your motto. Don't swarm the model with too much data. Focus on providing rich informative example. If you are not sure how many examples you need, start small and irritate. Keep adjusting and tasting until you find the sweet spot where the model performs well. Remember, the key is experimentation. There is no one size fits all the answer. I hope now you understand how many example you should be putting while predicting or clustering or maybe doing a data analysis on that. So if you found this to helpful, please let me know. And please also let me know how you're going to use this knowledge to work on your project or maybe you are going to use this knowledge to do daily task. That's it. That's the photos video. And I hope you are learning this concept about machine learning and all that, and I'm able to teach you in simple terms. So that was my main goal in this section. So I hope you learn that. Please let me know if you learn, like, anything. If you learn anything, please let me know. And that's it. That's the photos video, and I'll see you guys in the next one. He's out. 26. Using Templates to Make Prompting Easier in ChatGPT: I wish you could make your daily task easier and more efficient without breaking a Swett. Imagine having a magic tool that helps you filling the blanks, making everything from writing body guard to planning a trip a breeze. Well, today we are diving into something just like that. The template pattern in hag petting. Trust me, by the end of this video, you will see how this simple concept can transform the way you interact with technology. Hi, everyone hadnir. I have spent countless hours of exploring the amazing way we can use AI to make our lives easier and more organized. The template pattern can open up a world of possibility for you. So let's get started. Have you ever followed a recipe to bake your farete cookies? You have list of ingredients and step by step interaction, right? Well, the template pattern works in a similar way when you are using chat Ji pitting. It's like having a ready made framework that you can fill in with your own details, making task quicker and more consistent. Remember, when I first started using hat GPiDy I was overwhelmed by all the possibilities, but then I discovered the template pattern, and it was like finding a secret key that unlocks so many doors. Suddenly creating content, organizing information, and even planning events becomes so much easier. So what exactly is template pattern? Simply put, it's a way to create a structure with a placeholder that you can fill in with specific information. This make your interaction with age pity more efficient and tailored to your needs. Whether you are writing a letter, creating a lesson plan, or even designing a simple game, the template pattern can save you times of efforts. So in today's builder, we will explore three main aspects of template pattern. First one is basic template, how to create simple template with a placeholder, and second one is advanced formatting, adding specific instruction for how you want to information to appear, and third one is practical application. Like real life example of how you can use templates to simplify your task and stick around till the end because I will share some exciting tips on customizing templates to fit your unique needs. Let's king things up with the basic. Imagine you want to write a birthday invitation, but don't know where to start. A basic template can help you by providing a structure with blanks to fill in. So for this example, I wrote props like this. You are invited to place order name, a birthday party and also, again, I added place order and then also added location as a place order. Then join us for fun games and a cake. You know, like when I was to do coding in Python, that time we used to put some brackets. So when we will input, then that bracket can fill with that input. So we are doing exactly the same over here. If you like, learn some programming languages like Python, Java or JavaScript, then you might know this kind of concept. So this is our basic structure of template pattern looks. So over here, what you can do is that you can pass the name list of names. After that, you can pass the date and the location. So Cha J PT will automatically learn that, and it will create messages for all those people. I use a similar template when planning my nephew's birthday party. Instead of writing each invitation from scratch, I just fill in the blanks and send them out in no time. It was easy and kept everything consistent. Now that we have covered the basic. Let's dive a little deeper. What if you need your information in specific format? This is where advanced formatting comes into play. Now just imagine you have list of people, and you also have bird date, and you want to bird date in this format like date, like month and after that year. So so far this, you can write your template format like this. Like please enter date of birthday in a format like this one, and your initials. Maybe let's say min Bogari. So initial will be CP, something like that. By specifying the format, you guide Cha JEPD to provide the information exactly how you need it. This is especially useful for things like filling out forms, creating should or organizing data. For instance, if you are creating a weekly meal plan, you can set up templates like this. So for this example, I wrote like this. Monday, the placed is breakfast, then gin placeholder lunch, and gin dinner. So you have to mention what will be in this place folder. I know ChargePSmart, it will figure out on its own, if you give this what you want in that placeholder, it will not get hallucinated. It will not give you the false out. After that, I also wrote, create a weekly meal plan template with placeholder for breakfast, lunch, and dinner each day for a week. Let's if I don't give this prompt, this one, and if I don't provide this one, the breakfast, lunch, and dinna and if I just provided the ware boxes, in this case, hat Gibt will 100 percently hallucinate. But now, if I skip this part, then it will not get hallucinated because also mentioned over here, you have to give me breakfast, lunch, and dinner. But over here, you have to mention the placeholders. So then it will add that data into that placeholder. Finally, let's talk about some practical application of template pattern. That can make real difference in your daily life. Imagine you are planning a family vacation. Instead of starting from scratch, each time, you can use vacation planning templates. Okay, I hope you understand how template pattern works and how to add placeholder for your outputs. So pause the video and I told you what we are going to do in this video. Like, we have to plan family vacation. And so pause the video, do that. And if you're not able to do that, then watch this video from now. So from here, I'm going to solve this problem. Okay, I hope you've done that, and if you are not able to do that, then this willow from here. So I said, create a vacation plan template with placeholder for destination date activity and accommodation. And over here, I wrote some placeholders, so Chat JP will understand where it should be putting its data. And we can also mention some destination. So according to that, it will also figure out all the things. Another great use is for creating personalized letter or thank you notes. A template, you can easily insert the recipient names and personal messages, ensuring each note feels special without the hassle of writing each one from scratch. To wrap this, let's recap what we have learned today. We explore the template pattern and how it can simplify your interaction with ChagpD by providing structure frameworks with placeholder. We looked at basic templates for everyday task, advanced formatting for specific needs and practical applications that can save your time and efforts. Understanding and using the template pattern can make your life easier. Whether you're organizing information, planning events or just looking at streamline your daily task, it's a powerful tool that can bring efficiency and consistency to your work and professional projects. Now if you're excited to start using templates, here are a few next steps. Try creating your own templates. Start with something simple like rose list or daily schedule. Explore more advanced templates. Look into templates for specific project like event or planning or budgeting. So that's how template pattern works. Please let me know how you are going to use this method in your daily task. In resources, I add examples, try them, copy them, paste them, experiment them, and also add your own data. Try to make your own templates. That's a portal Vodo. I know you might be confused, like what is template pattern, how we supposed to do with this. So let me tell you in short, let's say you have huge amounts of data like Excel sheet or huge amount of paragraphs first past is over here until habit, like you can add template. Let's say you have a list of 100 people are coming to your party, and there are in that list, you have name. So, like the list, you want to invite them one by one. If you do that manually, it will take maybe day not day, but hours. It will take. While writing Tam Bland, you have to add placeholder and in that placeholder, add the name name of the guest. Like, create a square bracket and in that add name of guest and Ja JP. In this placeholder, you have to add the name guest name, and after that, you have to write a random invitation method. Or you can add one specific message, or if you want to create something unique or a unique message to each guest, then you can write at the end. You have to create a random message for each guest. It has to be unique. Everything has to be different, not same. So this is how you use a template pattern in real life. Now, I hope you understand how it works and where you should be using this tembldPattern. So that's the porto video. In next video, we are going to use we are going to learn mark Dos. It is also really amazing concept in hat Jeopardy, so Sayo for that. So that's what do, and I will see you in the next one. Is out. 27. A Quick Guide to Markdown Formatting in ChatGPT: A man student, have you wondered how you can make your takes stand out in Chachi Petty effortless scene? Well, today we are diving to magical world of mouth Downs. And trust me, it is going to change the way you interact with text forever. Let me share a quick story. A while back, I was working on a project and I needed to organize my notes. I tried using plain text, but it was just so boring. That's when I discovered markdown and everything becomes so much clearer and more colorful. Today, I'm here to show you how Modowns can do, same for you. So what exactly is Markdown? Simply put Markdown is a lightweight way to add formatting to your text. Think of it as giving chat Jeeperd instruction on how you want your text to look. Whether it's heading, bold words, list, or even links, and the best part, you don't need to take Visa to use it. So here we will cover today in this video. First one is basic formatting like heading and emphasis. Then second one is creating a list and links. And third one is adding tables and footnotes. We are going to explore these three main areas, and I promise you will find some cool tricks along the way. So let's get started. First up, let's talk about headings and emphasis. Imagine you are writing a story or report. You want certain parts to stand out, right? That's where heading comes in. In markdowns, you can create heading by adding few special symbols before your text. For example, one hash symbol makes a big heading, like the title of your story. Two has make the smaller heading perfect for the chapter or section. Oh so there is a one story, and let's say I want to apply markdown on it. So let's do some markdown and let's see how it works in hajbet. In Chad Jept I wrote like this. First one is hash. This one is title, so I add a single hash, and this one is like small heading. So that's why I added two hashes in that in it. Let's give this to Chi Jept and let's see what kind of food we get. So as you can see. So for this, we added one has, so we got the big title. Big title means the big heading. After that, for two hashes, we got small heading. And after that, it continues the story on its own. Because I didn't mention over here, you have to continue the story or you have to just do this. Still chargeabi continue with this. And when you want to emphasize something like making it bold or italic, markdown makes it super easy. Just wrap your words with asterk or underscore. For instance, italic or bold. It's like giving you words a little extra flare. Okay, so for bold, use two asters. So as you can see in this example, I use two asterix, and for Italic, I just use single asterisk. So let's give this two chargeb Okay, so as you can see, with our first this paragraph, we got everything in bold, and after that, the second paragraph, everything is in Italy. And after that, this one is a random story made by Chat Gibt. But we are only interested in this one. So as you gonna see, it works. Now let's organize our thoughts with list and add some cool links. Whether you are making a shopping list or outlining your weekend plans, Markdown has got you covered. So for simple list with bullets, you just start with each item with a dash and a sik like this. Let's say we want to make list of fruits. So I add three fruits in it, and I also added this dash before each item. And let's add it over here. So I'm performing Markdown's operation over here, so it will understand it has to do only this part, not has to produce random output. Okay, now let's give this list to hang bit, and let's see it works or not. Okay, so as you can see, it works, but it still mentioned markdowns, I don't know why. Okay. So it's supposed to look like this. Well, let's say, when you have document and in this, you want to mention some list. So just add dash in front of that item, and it will make that thing into list like this bullet list. And if you prefer number list, then you can just add numbers like this one, two like this. As you can see, this is pretty neat, right? Now, let's say you want to add a link to your Ferrot website. In Markdown, you do like this. Let's say when you want to add link or website like open AI, in this case, you can just add text for in the square bracket and after that, make rounded brackets and in that add the link. And it's become clickable link, handy for sharing resources or references. So this is how you add links in hareby if you have any document in that, you want to add specific link or something. So in this case, you can add link in this way, like this. H, Oxo output will look like this, but I don't know why it is showing the link as well. It's not supposed to do this. It only supposed to do like this. Let me show I have some examples over here. So as you going to see, it has to look like this. And this is link. If I click this, it will redirect you to that website. Finally, let's level up with tables and footnotes. This might sound a bit advanced, but trust me, they are easier than you think and incredibly useful. Imagine if you want to compare different fruits, here's how you can create a table. So table format, you can write like this. This is called Pipeline operators. If you are familiar with programming languages like SQL and all that, then we call it Pipeline operator in that. I used to use when I was in college, when I was doing engineering. And in Chat GPD, we can do the same. You can write like this. And let's give it to hatGPD and it will convert that into table. But still, I need to provide like this. Okay, output has to look like this. I don't know what is going with Chad Jeopardy has to look like this. If we provide your data in this format, it has to look like this. But let's say let's say you have data like this. So in this case, what you can do is that you don't want to use this pipeline operators, the um, lines. So in this case, you can just, like, type. So you can write like this, convert this into Taba format, and it will do the same thing. So as you can see, it is doing the same thing. Tables help you organize information neatly and makes it easy to read. And what about footnotes? They are perfect for adding extra information without learning your main text. You simply add a number in bracket like this and then provide the footnote at the bottom. It's a great way to add details or reference without interrupting the flow of your content. For footnote, this is the example. So this is where our text will come. Okay, first run it, then we understand how it works. Okay, so in output format, it looks like this. So as you can see, it is over here, and it carries the extra information we provided over here in this while writing the prompt. So when you click this, it will redirect you to this information like this. Okay, now let's take a final example. In this I ddt all kind of markdowns which are available in Chat Jeopardy. Like I took the example of exercise. Fst is data for title, that means the big heading. After that, the double hashes for like a medium sized heading. Okay, after that, we added two asterix for bolt and again, asterix, and this one is for Italic. And we also added a dash for the list, and this one is a table. So let's run this. I hope it runs. Well, okay, it's finally. I started a new chat. Now it's working good. Okay, so we have the four hash, double hash. Who, this one is the bolt, two asterix, and where is Italic. Okay, so here is a italic thing, single Asterix and this one is a table. And then we have a footnote, and it is redirecting to this information over here. And we also have here a footnote, and it is redirecting to this. And this is redirect to this. Wow, we covered a lot of today. Let's quickly recap what we have learned. Heading emphasis, make your text organize and stand out, list and leaks, keep your information structured and interactive. Tables and footnotes present data clearly, add extra details seamlessly. Markdown is a simple eight, powerful tool that can transform how you had GPT, making your interaction more organized and visually appealing. Whether you're student or professional or just someone who loves neat text, Markdown has something for you. If you are eager to dive deeper, check out the resources in your description below. Okay, so in resources, I add two links, and first one is Github repository. And in this if you want to learn more about how markdowns work, um, there are there's huge amount of data over here. You can learn from it, as well as there is a second link, and from here, you can learn as well. The same thing, how markdowns work and extra markdowns as well. It is over here. We can also make checklist in using markdowns. We will see that in Anons markdowns video, so stadium for that. So that's it. That's how Mark Dom works. And industries also added a few examples, copy them, paste them, try to add your own datas, permit them, also added assignment, do that after completing that project section. So that's it. That's the photo today. I hope you understand how Mardom work. And I personally used it like there was on Google document, used the Google document, and there was just a text. And I wanted to make it more, like, visually appealing. So that time I used markdowns. And I know it is a bit hard to understand at the beginning, you might be saying, we can just add like text, make a list or make this into bold or it a if there is a huge amount of data, then it will be unpractical or maybe it will take too much time. So in this case, you can use mark dolls. So that's what Dsudio and I'll see the next one pets out. I think 28. Verifying Facts and Staying Accurate in ChatGPT: Hi, everyone, hedanir. Today we are learning about self consistency, fact checking, and referencing footnotes. And basically, we are trying to get accurate results from Chat Jeopardy by providing data, and it will not hallucinate at the end. It just meant we are getting accurate data. Mostly, this video is theory based, and at the end, we will look at one example like how we saw the exercise example. It is exactly like that. So let's start this video. Have you wonder how Chat Jeopardy keep its information accurate and trustworthy? Today, we are diving into a fascinating world of self consistency, fact checking, and referencing footnote in ha Jebedi. By the end of this video, you will see how these tools can transform the way you interact with AI. Let me share a quick story. Last year, I was helping a friend write a school project using Cha Jeopardy. At first, everything seemed perfect, but then we noticed a few hiccup with the facts. That's when I discovered the magic of self consistency and footnotes in ha Gibi. Trust me, it's changed the game. So what we will cover today. In this video, we will explore key areas like self consistency in ChargePi, the importance of fact checking, and third one is how to use footnotes for reliable information. So stick around because we have some amazing tips and real life example that will make you hagiPiP. First up is self consistency. Imagine you are telling a story to a friend and you keep the details straight throughout. That's exactly what self consistency does for Chad Gibt. It ensures that the information remains uniform and reliable across the conversation. For example, if you ask Cha Jept about the word cycle today and then again tomorrow, self consistency make sure it provides the same accurate information each time. This builds the trust and makes interaction smoother and more reliable. Now next, let's talk about fact checking. Just like you won't want to believe everything you read online without verifying ha Ji Bri uses fact checking to ensure the information it provides is accurate. Say you ask, what's the tallest building in the world? Ja Gib not only gives an answer, but also checks it again reb source to make sure it's up to date. This way, you can trust the information you receive. Finally, let's explore referencing footnotes. Footnotes are like litter helpers that show where the information came from. They add the extra layer of trust by putting you the original sources. Imagine you are writing a report on climate change. Cha gibt can provide detailed information and include footnotes that reference specific studies or reputable articles. This not only backups the fact, but also make it easier for you to verify and explore further. Okay, I hope you got the theory about self consistency fact checking and referencing footnote. So now let's take in real life example. So I have some data about coffee, uh, so as you can see on the screen, some data about coffee, and after that, I want to ask. First, I want to format it in using markdowns, and after that, I want to fact check as well as, I will add foot rolls for the references. So let's give this to Chat GPT. Finally, we got our output in this format, and as you can see, we wrote first giving facts about coffee and first one title. And after that, small headings, and we added, like, fact. Also, I added fact because Chat Jeopardy was hallucinating. It was creating output like this. So that's why I added fact inside it, and then it gave me output like this. Okay, so I don't know why Chat JEPD is being weird today, but, but then we got output. That's what's matter. Okay, so we have uh, footnotes and as well as the referencing footnotes, as well. So if I so these footnotes contains accurate information. So let's say if someone clicks on it, so it will redirect this through this information. And as you can see, Cha Jeopardy is broke, I have to repress it again. And now if I click on the third, it will redirect me to the third point. So if someone tells you had Jeopardy is useless, it always hallucinate. It will hallucinate. But if you provide the data accurate data, it will not. And if you wrote a prompt in such a way like this, we wrote a prompt like this. First, we gave the information about it, then we told we want information in this format, as well as we told you have to reference the footnote, then it will not hallucinate, and it will not create ten data from anywhere. And as you can see, it's also creating self consistency over here by using these footnotes. I hope you understand what we are trying to do in this video. Okay, now let's recap what we have learned in this video. First one is self consistency, keeping information reliable through Apncon. Fact checking is ensuring the information provided is accurate and up to date. And third one is referencing footnotes. Avoiding trust by linking to original sources. Understanding these features make your interaction with ha GP more effective and trustworthy. Whether you're a student, a professional or just curious, these tools can help you get the most out of AI. If you want this helpful, please let me know. Okay, I want to tell you one principle. It's called PoitoPrinciple. It is also called 80 20 principle. So, whatever you learning from this course, you are going to use 20% of the thing you learn from this course. So I know this topic is relevant for some people, and it is not relevant for you might be. So if it is not relevant, then you can totally skip this and you can focus on the technique, like let's say, you want to learn you use in context or learning a lot. Then master that, you don't have to master this one because this is for writing articles or creating accurate information more like a research purpose. But what I found is that markdowns are really helpful. Let's say you have huge touring or a huge article about something and you just wrote it in Google Dot. And now you want to convert that into accurate format. So in this case, you can use markdowns. So remember, you are going to use 20% of the things, what you are learning from this course. It applies everywhere. I literally mean it, it applies everywhere. So that's what do I hope you understand how it works and also learned about product to principle. So that's what do and I will see you in the next one piece out. 29. Advanced Markdown Techniques to Enhance Your Prompts in ChatGPT: Hi, everyone here. Today we are learning about more advanced mood on templates. And, you know, you can create a checklist in Chat J Pet using Mack Downs. Like, I use producty apps a lot like Tolist. And if I saw the wall in front of me, there is a huge sticks, and as you can see, it's OE as well. I created lots of taskfming and I use kind of check list kind of thing. So I was like, experiment with Cha JP. Can we use markdowns in Cha J Bid to create same thing? And so I can just copy that thing, copy that list and I can pers into notion. So so in Notion, I don't have to create a list again and again. So that's when I discovered this technique. Now you might be thinking, Sathan what exactly are Advanced Markdown templates? And why should we care? Simply put, Markdown help you format your text in a clean and structured way, making your chatty play experience smoother and more effective. Today, we will explore some exciting ways to use these templates, including creating interactive to list, beautiful newsletters and even personalized journals. So first up, let's talk about interactive To Do list. Imagine having a personalized check list that you can update and manage right within Cha J Peten. It's perfect for keeping tracks of your daily task, planning a party, or you're organizing a family outgoing. Few weeks ago, I bought this book called like an artist. So in the end of the topic, I guess there is a thing. Okay, in ninth chapter, there is one topic, and the author says you have to list down everything what you do in, like in daily life. So in his book, like, he shared his logbook, and in this, he wrote Ps plus Indian Lefto. So he was like journaling or, like, writing what he did in his day. And he said, we should do this like more often because this pas of creativity is something and something like that. And we can revisit to refresh our memories. I thought let's give it a try and let's create a list, you can do it manually as well, but sometimes I like to keep everything digitally because if I write that in notebook, I'm thinking someone will read that. That's why I'm to store it digitally by using Notion or maybe in Lord's app. Okay, now let's see how you can do this in chat hippity. Here's how it works. You start with a dash and empty brackets like this. And then add your task, for example, buy groceries. Aji Betty can help you create a detailed to do list with category, deadline, and even priorities. Let's say you're planning a weekend trip. You can create section for packaging, travel arrangement, and activities all neatly organized and easy to update. So let's create to do list for that. So I wrote like this. For tools, we have dash and were brackets like this. And they add you item like this. And if I give this to hit JBT, it will create two list for me like this. But in output, hGPT added some few details as well. I don't know why. I just wanted this thing and the boxes, but it is okay. Now we can copy these things. Okay, now I came to Notion and I can just paste it over here. And as you can see, we got list, I can just check like this. Okay, selected. I can check each item like this. And when I was practicing this technique, so I created this. So as you can see, how cool is that? Now, I want to hear from you what one task you had love to organize with to do list. Drop your ideas in Commands below. Now, let's explore how to create beautiful newsletter using Markdown template. Whether you are running a club, sharing update with your friends or just keeping the personal journal, a well formatted newsletter can make your information shine. With Markdowns, you can easily add headings, bullet points, images, or links to your newsletter. For instance, you can start with a catchy title like this, hash weekly update, and then add section like highlights, upcoming events, and fun events. Then ha Jetty can help you format everything perfectly, making your newsletter look professional without any hassle. As for this example, I wrote this kind of prompt like a first news weekly update for title, then subtitle headings and some information in it, and then we got output like this. Now finally, let's dive into personal journal. Journaling is wonderful habit for reflection and creativity, and with markdown template, you can make your journal entries more organized and visually appealing. We can start a date and heading like this. Two hashes after that, April 27 and 24. I'm just taking in your random date, then add your thoughts, experiences, or even sketch using simple markdown syntax. You can include bullet points for daily highlights, bold text for important notes, and even links to your favorite course or resources. So pause the video and try to complete this problem on your own, add markdowns to it, and then we will compare our prompt. Okay, I hope you've done that. So this is how my prompt looks. First one is April heading. Sub adding, then again, add a heading, then attend the fest workshop. And after it and at the end reflection like this. So this not only makes you a general entries more engaging, but also help you keep track of your personal growth and memories in structure way. Plus, it's super easy to update and customize. I know in this video, I did not take huge examples, but the main purpose of the video was to teach you how to use markdowns in real life. So in Markdown video, I give you the two links to resources. I hope you visited that links. In that, you will get lots of markdowns. I hope you tried that. If you're not, please go to these resources again and copy those markdowns and try to experiment them and try to make something from it, try to format your data in a different way. You know, I was telling you, like, you can do the same thing without writing markdowns. You can write the same thing in text format as well. So if we take our to do list example for this, you can write problem like this, create an interactive to do list for planning weekend trip, including categories for packaging, travel arrangements, activities, use Markdown syntax for formatting the list. Send us to format the list. Okay, it gave us the bunch of points. You know, had GIP, I don't know. Today is hallucinating a lot, but previously, I tried the same thing. So when I was experimenting with this technique, um, I gave it in text format, the prompt. I didn't mention the markdown language like this. Still it gave me the result like this, but I don't know it is hallucinating war. But I don't put output like this. But I can mention it over here. I want in something else. I don't like this, something like that. Okay, let me show you one thing in Chat Jpeti. So let's see, let's say. Let's say you want to create a list. Okay, I wrote one prompt, really weird prow created list of random sums. So when I execute it, it actually shows the markdowns. So if I run it, you have to watch it really carefully. I will also point it out. So, okay. I created very fastly. So before executing it, it actually writes down the markdown, then convert this thing into list like this. Okay, now let's take another example. I really want to show you how it works behind the hit GPD. When we upon it GPT, it just situate the markdown and instantly create the actual port. So let's see, and I hope I'm able to show you the thing, what I'm saying. Like, just look at when I run it, just look at the white dot and then you will notice the thing. Okay, I ran it two times, and it was so quick. I will slow down the video, then you will notice, there was stars like two stars before running this. So if I run this again, you have to watch it really quickly. Keep. So just focus, okay? Send. Okay, so as you can see, there was two stars. When I used to do this, it was like slow, but now it becomes so fast. So that was what I want to show to you. Wow, we have covered some fantastic way to use Advanced Markdown templates in Ja JBT today. To recap, we explore interactive to do list to keep your task organized, beautiful newsletter to share your update stylissly and personal journs to document your journey creatively. These tools are not only easy to use, but are incredibly powerful helping you make the most out of your Ja JBD experience. Whether you're student a professional or just someone looking at to stay organized, marked on templates have something to offer you. If you want to dive deeper, check out the links in the resources for more morbdn tips and templates. So that's how you can use Mark Dows in real live and how you can create checklist. So that's about judo. I hope you understand how to use this technique. And in a stadium for next video in next video, we are going to learn some important thing. So stadium for that. That's about Vudo and I'll see you in the next one sub. So, in. Is. Is Don't Don't 30. Escape Strategies: Handling Errors and Blocks in ChatGPT: Hi everyone, Chinre. Today, we are learning about escape values in prompts. If you're a programmer, you know this concept like EL. In programming we write this condition if this condition is true, then exhibit this. If this is false, you don't have to exhibit anything. So we are kind of doing similar in prompting today. So let's start the video. Have you filed first at when you are trying to fit something into process and it just doesn't work out, no matter how hard you try. Well, in today's video, we are going to explore a cool little trick that can prevent this very problem when using chargeabiPmpt, something called escape values. By the end of this video, you will have a whole new perspective. How to make Jajibi follow instruction better and what to do when things don't get as planned. So a while ago, I was working on a prom for fun quiz game. For my student, I gave Jajibi a list of questions. You struggle with quotient that didn't have straightforward answer trying to fit a square peg into round hole. That's when I learned about escape values. A way to give hagiPit an option when the situation just doesn't fit. And, believe me, it made all the difference. So in this video, we will uncover what escape values are, how they work, and how you can use them to get better and more accurate results from chargebty. By the end of this video, you will know how to avoid these frustrating moments and take control of your outcomes. So are you ready? Let's dive in. All right. Let's start by explaining what escape values actually are. Imagine you ask ha Jib to provide a list of animal limbs, but in your prompt, you accidentally ask for color instead. You know ha Gibi is clever. But when it gets to trying to follow instruction, that doesn't make sense. Like asking for a color from a list of animal, I needed a way out. It will hallucinate and it will create random data. That's where escape values comes in. It's a simple technique where you provide the AI with a backup plan. You are basically saying, If you can't do what I'm asking for, here's an alternative. For example, instead of asking the color of animal, which may not always exist, you could ask if there is no color, right, not applicable or NA. Let's say you are working on a prom for recibe guide, you could ask GBD to list all the ingredients and amounts, but what if some ingredients don't have the exact measurement? Instead of leaving it blank or trying to force a random guess, you could use it escape value and tail it. If no measurement is available, just write as needed. Okay, let's take an example of animals. So when we was discussing the first example, we took the example animal, but I forgot to execute that problem. So I ask Chat deputy like this. If no color is associated with the animal, respond NA and shave. Or you can also use false or no. You can write anything. So let's say, in this case, goldfish, does not have color mentioned it over here. So if I give this, it will mention NA like this. So if you don't mention NA, it will try to get data from anywhere, and maybe it will hallucinate and at the end, you will get inaccurate result. So that's why I add NA or escape value into it. A few months ago, I went to one restaurant nearby house, and I ordered zucchini noodles with pesto. And so I try like I was trying can recreate that dish in in my house, and I asked AJPti what will be the ingredients. So Jepti give this kind of recipe and ingredients. So let's say, let's say we don't have this olive oil, one cup let's say we don't have the measurement for olive oil, and then I can write like this. If the measurement for the ingredients is not available right as needed, so I'm not going to add too much oil in that. I will add the oil as much I need. So let's see it. So as to see habit got it. Olive oil as needed. Now that you know what escape values are, let's look at how they can prevent confusion. When you give Cha GP a task, it's going to follow your instruction as closely as possible. But what happens if the instruction does not fit the data it has? Well, without an escape value, it will often try its best and that can lead to incorrect or confusing result. By including the escape value, you are preventing it from making awkward guesses or providing wrong answer. This way, the results stay neat and relevant. Now let's move on to the most important part. Why this matter for you, Calu can completely transform how effective your proms are. Think about this. Without them, you are leaving hagiPD in situation where it tries to force answer into format that might not work. This can lead to results that aren't useful or worse. For example, imagine you are making a list of historical events. Some event doesn't have specific dates. Instead of a gibt filling in the data that might be inaccurate, you can give it a escape value. So you can write prompt like this. If no data is available, write date not specified. This simple addition prevents strong information from creeping in and ensuring your final output is accurate and clean. Escape values are like a safety net. They ensure that even when the data doesn't fit perfectly, the output remains unstable and you don't have to worry about incorrect or forced information. And that's the magic of Escape value. To recap, Escape value give age By a clever way to handle situation where you are instruction just not quite fit. They prevent confusion, ensure accuracy, and help you get better result. Whether you are working on a fun quiz, a family Triva game or a project with missing information, escape values are your go to solution for those tricky movement. I hope this concept help you in your next prompt creation, and don't forget if you're curious to learn more about Advanced prompt technique or want more tips like this. Lastly, if you want to dive deeper, check out upcoming videos where you will explore more advance prompting your technique and how to make them even more dynamic. Ogindsource is added example, try them, and also add a assignment, do that, and sim posization. That's a four do video. I hope you understand how to use Escape values, and it is easy. We just have to add a Senn that's how it works. So that's for video, and I will see you then the next one. Is out. 31. A Beginner's Guide to RAG (Retrieval-Augmented Generation) in ChatGPT: Imagine asking you AI question and receiving perfectly right or accurate answer is time. Sounds amazing right. But what if I told you without the right information, you were smartest EI could just guess and get it totally wrong. I once asked HGP about the best restaurant in small town that I visited recently. To my surprise, it confidently gave me an answer. But when I checked, the restaurant it mentioned didn't even existed. This got me thinking, how can we make sure EI doesn't just make things up but gives us the right information. This is where something called ratable argument generation. In short terms, it is RAG comes into play. Today, we will explore what RG is, why it is important, and how it makes AA smarter by fetching the right information before generating an answer. We will dive into why AI sometimes give the incorrect answer and how RG solves the problem by retrieving accurate information and how it can improve the quality of AI responses. Stick around because by the end of this video, you will have a whole new perspective on how AI can be more accurate and reliable. Let's start by understanding why AI like Cha GPD can sometimes give incorrect answer. Think of it this way. When you ask EI a question, it tries its best to provide a response based on what it already knows. But here's the catch. It doesn't always know everything. For example, imagine asking Cha JEPD about who won the very recent school chess champion in your neighborhood. Unless that information is widely available online, it may not have access to the exact answer. So it might make a guess, which could lead to mistake. This is called hallucination. When AI makes up an answer, that sounds correct, but it isn't. If I ask had Jibety like who won the 2024 local chess tournament in my town. So let's see, and we will also search on Google is accurate or not. So as you can see, Chat IPED is totally hallucinating over here. It doesn't even mention the name of that person. It is saying there was a tournament and there was a prize of 71,000 rupees. And I wanted to know the name of that person, but it did not mention it hallucinated because it doesn't have the latest local data. This is where the need of RG comes in. Now, let's talk about how RAG fixes this issue. Think of RAG like giving your AI a new tool. It's like handing it a library card, so it can go and fetch the right book to answer your question. When AI uses RG, instead of guessing it perform a search, it goes online. Retrieves the most relevant information and uses that to give you an accurate response. Let's say, you are curious about when the next nature photography event in your city is happening. With RAG, EI could go online, find the event schedule from the trusted source, and then give you the exact dates. No more guessing and just accurate information. Okay, so I asked this question to Chad Jeb, like, what is the next nature photography event happening in Mumbai? And it gave me the date. Date is October 9 22024. And I Googled the same thing over here, and it says it is happening between October 5 to January 5. So as you can see, it is not hallicinating over here, but it gave them, like, information. And it also referred from six sites over here, like wild photography of the exhibition. And this is the same source materal used by Google. So if we look at it over here, NMSCC, as you can see NMSCC, it used the accurate information. And here's the most exciting part. RIG can dramatically improve the accuracy of EI in areas where specific up to date information is critical. Imagine a scenario where a teacher needs to know the latest government card lines for student re opening. Without RG, the EI might give old or incorrect data, but with RIG it can retrieve the latest policies from official sources, ensuring the teacher it has the most accurate and correct information. This makes AI a powerful tool, not just for general information, but for real time specific answer. So what did we learn today? A is amazing at answering question, but it's not perfect. When it doesn't have the right information, it might guess and get things wrong. That's where RG comes in. By retrieving the right data, it makes AI much more reliable and accurate. RAG is crucial if we want AI to be more than just a guesser. It's how we make sure the answer we get are backed by real accurate information. You're a teacher, a business owner or just someone curious about the world. This is the big step for making AI even more useful in our everyday lives. So before we end this, I want to tell you one thing. So when I started using Cha JBT that time, it didn't even had the Internet search, like, as you can see over here, so now it goes to the Internet, its searches, and then give us the answer. That time it just to say, my data is up to 2021 and I can't face the real time data. So with time ha JBT model will get more accurate and it will give the information accurately. I hope so it happens. Let me show you one thing. So when I was, like, giving the example of I searched for the best restaurants in my locality. So I did the same thing over here. I added the prompt over here. And, um, in this, it says, Hotel Boy Bobi. But if I google this, as you can see, um, the hotel Boy Bowie is not even in my area. Hotel Boy Bo is there, so it is totally different. And, and there are some other hotel names like hotel. This one SudiEecutive, and I think this one is there. Let me check once again. Okay, so as you can see, it is over here. So it's kind of hallucating as well and giving us the not accurate, but giving us the information, which are up to date. So that's all about RAG. I hope you understand what we are trying to do here and how charge work and how RAG work. So this is what will do, and I will see you in the next one. 32. Retrieval Methods: Using Search, Databases, & Embeddings in ChatGPT: Hi, everyone Ciner. Today we are exploring something exciting, like how Chat GBD retrieves information from data voices, embedding and searches. You might think, wait, isn't that a bit complicated? Don't worry. By the end of this video, you will see how simple and fascinating this can be, and it might just change how you look at AF forever. Just the other day, I was trying to find an old recipe my grandma used to make. I had no idea which notebook it was in, but I remember a few key ingredients. It felt like researching for needle in a haystack. But imagine if I could have simply asked an AI to retrieve it for me. That's where today's topic comes in. How AI like ha Ji Pity searches for the writ information. So what exactly are retrieval approaches? And why are they important? Habit, retrieval is about finding the right information from a large pool of data, whether it's searching a document or pulling info from the database or using something cool called embeddings. Retrieval is how Cha GPT gives you accurate and relevant answer. In this video, we will cover three key ways hagibRtves info, search databases and embedding. By the end of this video, you will understand not just how Cha GPT finds answer, but how you can take advantage of this process when interacting with AI. So let's dive in. First up, let's talk about search. This is the simplest way HGPT retrieves information. Think of it like using search engine. You type in question and it looks for relevant piece of text to cue and answer. Let's imagine you are looking for a quote from your favorite book. Instead of flipping through pages, you ask ha gibt and searches its memory to find exactly what you're looking for. It's like having a personal librarian at your fingertips. Search works by scanning through data to find the right keyword or phrase, just like we might scheme a book or an article, but hatibiy does it much smarter and more efficiently. Now let's move on to the database. Databases are like super organized filing cabinets where information is neatly stored and labeled. Imagine you are teacher trying to find student grades from the past year. Instead of flipping through the stack of papers, Ja Gibti can pull that info straight from the database. It's designed to retrieve precise structure information quickly. This is really useful for organization that relays on detailed and stored data. Picture this. Ja chibit access a database of all the books in library. You just ask which books have won awards in the last five years? In seconds, Jagibt retrieves a list. No more searching shelf by shelf. And finally, we have the secret sauce embeddings. This is where things get really cool. Embeddings allow Cha Gibt to not just search for exact words but to understand the meaning behind them. It turns text into numbers. Think of it like translating words into cornars on a map. Text with similar meaning are placed closer together, making it easier for Cha Jibe to find delivered information, even if the wording is different. Let's say you ask a jibe, what's the good way to stay healthy? Instead of just searching for those exact words, hagibi understands that exercising, eating well, and sleeping enough are all related to staying healthy. Thanks to embedding, it retrieves a wide range of helpful tips, even though you didn't mention them directly. It's like having an AI that thing beyond just keywords. So to recap, we covered three powerful ways hagibi retrieves information like simple search where it looks for keywords and database. Databases where it finds structure information and embeddings where it uses the meaning of text to deliver even better answers. These approaches make Cha Gibt more than just questions and answering tool. They make it powerful assistant that can pull from the vast ocean of knowledge and find what most relevant to you. If you want to dive deeper into how ha Git works, I will add the links of some resources or website so check them out and go through them and learn how ha JBT works. So there's a portlD and I hope you understand how haiPt works behind the back end of hag Bitty. So there's a port judo and I will see you as the next this. 33. Boosting Results with Prompt Engineering for Augmentation in ChatGPT: Imagine you are baking a cake, and start adding sugar, you add salt. No matter how good the rest of your ingredients are, the end result will be far from delicious. Actually, this happened me a while ago when I was in hostel. I was trying to make something, and we added some other ingredits in. And when we had it, then we realized that we had something else. Just like baking getting the right ingredients, in this case, the right information into a prompt is crucial when working with AI. When I first started using AI, I thought all I had to do was giving it more information and the result would automatically improve. But I quickly learned that's not always the case. You can give EI all the data in the world, but if the prompt isn't crafted carefully, things could go very wrong. Today, we are talking about prompt engineering, a fancy way of saying how we carefully design prompts to guide EI in the right direction. Specifically, we will dive into why this is so important when you are augmenting the prompt with additional information. If you have ever wondered why some AI responses are spot on, and others totally miss the mark. This video will open your eyes to the magic behind the scenes. Will cover some common pitfalls, real world examples and simple technique you can use to avoid mistake. So stick around because by the end of this video, you will have clear understanding of how you can control the output of AI, even if you don't control all the data it drives. Imagine you were asking hagibi for the name of the winner of the local spelling bee competition. You could either rela on the AI to guess or you could provide the name of the top three contestants and ask you to figure out who won. Let's say if I give prompt like this, who on the Springfield 2023 spelling Be. Okay, we got the answer, and this is the answer the literacy thing, I guess, I hope I'm pronouncing it right. And I'm not checking the information right now. It is accurate or not, but let's search on Google, the same thing. And Google is saying the Shah. Let's say if I type like this, who on the 2023 spelling be? It might give accurate result, but sometimes it will hallucinate, and it will give the totally random name. So okay, I just researching and Dev Shah was the winner. But let's say, but if I provide the top three contestants in this list, Now I have argumented the prompt with key information, making it easier for AI to get it right. So as you can see, in this list, it says the Sha was the winner of 2023 National Spelling Bee, and Haran Logan won the 2022, and Bruhad Soma did not win in either year, but he won in 2024. So this shows how important it is to control what goes into prompts. Simply giving the EMO data doesn't always improve the answer. It has to be right data. And here's where things can get tricky. What happens if the information we add it is unhelpful or worst or completely wrong? Imagine I copied the list of pokes instead of persistent names and added that to the prompt. Now, instead of helping, I have confused the AI, where it tries to figure out the winner. So it might say the 2023 spelling bee was won by Unluckly champion Intermenzo by Sally Rooney in surprising terms of event, the book emotion, so as you can see, it is totally hascinating. And it gave the like really weird output. This happens because the prompt was engineered poorly. If the information doesn't relate it to the task at hand, the AI can't make sense of it, leading to weird or completely wrong answer. Now, let's look at how to avoid this issue. The key is to structure the prompt in a way that makes it easier for EI to reference the writing formation. Let's say you are asking for details about historical events. You provide a list of key data and ask which date was the treaty Welius signed. I know I'm totally pronouncing it badly, but we got the point. So we will provide the list and then we will ask the prompt. By numbering the dates and telling EI to refer a specific number in its answer, you are given the clear road map. This makes it much harder for EI to get confused or make things up because now it knows to look at the correct information. What fascinating is that, even if the EI doesn't know the answer, it will recognize what it can find supporting evidence. That's huge win because it stops the EI from just guessing and hallucinating. Let's say if we don't have information about this sign thing, we can write it over here. Let's say you have a historical data, like 2000 lines of data, and you want to know this exact thing. And in this list, you don't have this data. So you can write it over here. If the data is not present here, tell me data is not over here, or you can also ask GBD, like give me some links, provide me some data so where I can find that information. By doing this, chat Bv will not hallucinate and it will at least give you the resources to look out. To wrap things, here are the key takeaways. Argumenting prompts with more information can be indirectly powerful, but only if the information is accurate and relevant. Poorly structured prompts can lead to confusing or incorrect AI responses. So it's important to think carefully about how we present the information. And simple trick like numbering facts or explaining guide the EI to use specific piece of information can significantly improve the accuracy of its responses. Prompt engineering is like being the director of the movie. You have to guide the AI, give it the right script, and sure it focus on the task at hand. It's a skill anyone can learn, and it makes a huge difference in getting the best possible result from AI. If you found it helpful, please let me know and let me know how you are going to use this. And I know you also face the same issue while prompting something. So let me know what you did that time and what you will do now you have the information to direct your prompt. So that's it. That's photo's video, and I'll see you in the next. 34. Overcoming Retrieval Issues: Noise, Size, and Relevance in ChatGPT: Have you ever tried to find something in huge piles of papers and only to realize you are looking at something different or maybe looking at something small? Well, that's little like how retrieval works in AI. We expect the system to find the right information quickly, but it's not always that case. Today we are going to explore some common challenges with retrieval in AI. Focusing on finding the right information, determining the right chunk size, and avoiding noise. This will give you the new perspective on how EA searches for information and why things might always go as planned. I remember once I tried to find an old recipe in stack of cookbooks, I found part of recipe on one page. But the key grains, that was on somewhere else. This is just like what happens when AI retrieve chunks of information. Times it wraps a piece that doesn't fit together or aren't complete. By the end of this video, you will understand why chunk size matter and how noise can creep in and how all of this impacts the EI ability to give the solid answer. And don't worry. I will break it down with small and fun example along the way. Let's start with the challenge of finding the right information. Imagine you are looking for instruction on how to assemble a bookshel, but instead, you get the page talking about different types of wood, close, but not quite what you need, right? This happens when AI reduce chunks that are similar but not exactly relevant to the question you asked. For example, let's say you are looking for information on how to grow tomatoes. The AI might bring back chunks about garden soil or fertilizers, but completely miss the step by step guide you actually wanted. Even though these pieces are related, they are not right answer. This can be really frustrating. Just like finding the wrong part of an instruction manual when you are in the middle of building something. Now, let's move on to something called chunk size. When AA pulls information, it divides takes into chunks. But how big should this chunk be? Should there be a paragraph or sentence or just a few words? It's like cutting a pizza. Do you want a big slice or small one? Too small, and you will miss the full flavor. Big and you miss you might get overwhelmed with too much at once. For instance, if you ask about movie plot and AI gives you the whole page of script instead of just summary, it's going to be tough to find the exact information you are looking for. On the flip side, if it gives you the only one line, you might miss the main point. Striking the right balance is key, but it's not always easy. Imagine you are looking for a specific rule from a game manual. If the AI splits the text into tiny chunks, it might give you the one sentence that say rule that die. The next sentence which says to move one piece forward, and it gets lost in different chunks. Now, you are left wondering what to do next. In Airtvs noise refers to irrelevant information that gets pulled in with good stuff. Think of it like searching through a drawer for pen but finding all receives. Rubber bands and everything except pen. In EI, this happens when system reduce too much unrelated content, making it harder to focus on what actually matters. Imagine asking for advice on how to take care of house plants and gives the chunk about indoor lighting for art displays. It's not completely undilated, but it's not helpful either. The noise clutters the result, and suddenly you are sitting through irrelevant details to find what you need. So what did we learn today? When AA detces information, it's not always perfect. It can struggle to find the right piece, give you the chunk that are too big or too small, and sometimes it pulls in irrelevant noise that distract from the real answer. But understanding these challenges help us get better at crafting prompts and working with AI to get more accurate result. Next time you ask a question, you will know that there is a lot of going on behind the scene to find that answer. That was all about issue with retrieval in a Gibt or in LLM models. I hope you understand this theory accurately, and now you know what should you do while writing prompts and how much information you should give and you should be giving accurate information not should be giving lots of chunky data. And if you give that amount of data, had GBD eventually hallucinate or will produce you weird outputs. So that's what's all about issue with travels. So that's about today's video, and I guess this is the last video on the AosPmpt engineering. And in next video, I will be sharing, like, what is the most important things you should be taking from this course. So that's the photo today, and I will see you when the next one is up. 35. Advanced Prompt Engineering Key Takeaways & Final Thoughts (enhanced): Everyone, thank you for sticking with me through this journey of Bras prompt Engineering. By the end of this video, I will promise you will live with a whole new way of thinking about how you work with EA tools like hat JBD, Cloud or Perfexlity or at the end, Google Germany. Today, I want you to give some key insight and techniques we have learned so you can walk away with confidence to try them out on your own. And I also know that you might be trying those techniques on your own to do prompting. Okay, let me tell you one quick story. When I first started learning prompt engineering, I remember feeling a bit overwhelmed, wondering if I could really make the air understand exactly what I wanted. But over time, I found that with a few techniques I could get some truly amazing results, and I want the same for you. So let's dive into these advanced steps together. Alright, let's go over some of the most powerful tools in our prompt engine tool kit. First step, one of the simplest but most effective technique is in context learning. This technique is all about guiding the AI by giving it clear examples instead of lensy instruction. Imagine you are trying to explain how to paint a picture. Sometimes showing a few example is all the artist needed, needs to get it right. The same goes for AI. Say you are working on getting the EI to write friendly email response, rather than explaining the detail, provide a few sample resources. The AI starts to understand the style, the tone, and you nuances based on the examples, making it much easier, get the responses you want without too many twigs. Move you on to another bowerful technique. It is rival refinements. This method is great because you don't need everything perfect right at the start. Think of it like sculping. You start with a rough shape. Then you keep refining it until it just write. For example, if you're crafting social media post, with a general bound. When the AI generates an output, pick the part you like, and then ask it to refine based on those. Each of iteration will make it closer to what you are aiming for. Almost like teaching a new skill step by step. And now here's a hidden gem that can make a huge difference. Whether it's in bullet points, tables or specific section can make all the difference. For instance, if you're using EI to generate reports, specifying a template a front like having the section for introduction, data analyst or conclusion help EI to stick to predictor structure. It's amazing how much cleaner and organized your result can be just by including these details. And last one is one of the most exciting relevation for just how vertatile these large language models can be. They are not just about chatting. We can apply this to all sort of classical EI and machine learning problems from classification to sentimental analysis and beyond. Tapping into these capabilities, we can solve wide range of challenges often more efficiently than building custom models from the scratch. And of course, we can continue to leverage these techniques. We have learned like incontextar iterator reforments to make those model work for us in powerful ways. So to wrap it all up, here's what to remember. Incotext learning lets you guide the EI with examples. Iterative reforans allows you to get closer to what you own step by step. And third one is output formatting is like roadmap, helping the EI deliver exactly what you need. The techniques, once you get the hang of them will save time and give the powerful results. And I already told you about Poreto principle, the 80 20 principle, the 20% things make you the 80% result. In this course, you are learning 100% things, but you will be applying only 20% of them, maybe less or maybe more. I hope you found these tips are valuable as I do. If you want to dive deeper, there are some resources I added in resources and make sure to check out our upcoming courses or classes on Mid journey and new futuristic AI tool that will help you in your work, as well as in daily life. So keep practicing, keep experimenting. And most importantly, keep having fun with it. The more you engage with these models, the more you will discover, and don't forget to share your insight and successes with the community. Who knows your contribution might just inspire the next big breakout in prompt engineering? Oh, thanks for joining me today. If you're hungry for more, be sure to check out our upcoming courses. So that's it for today's video, and I will see you guys in the next one. I know this was a thank you video, but there will be one outro, so please watch that. So pi now.