ChatGPT Prompt Engineering: From Beginner to Advanced

1. Welcome and Course Overview: Welcome to Chat GPT Prompt Engineering, the Ultimate Guide. In this course, you'll learn how to craft powerful prompts that optimize AI responses for various use cases. Whether you're a professional student or AI enthusiast, this course will equip you with the skills to get the most out of hat GPT. This course is structured to take you from the basics of prompt engineering to advanced techniques. You'll learn how to structure prompts effectively, troubleshoot AI responses, and leverage hat GPT for practical use cases. Expect a mix of theory, hands on exercises, and real time demos in this course. To maximize your learning, actively participate in exercises and demos. Experiment with different prompts, apply concepts to real world scenarios, and don't hesitate to test AI capabilities. The more you practice, the better you'll become at crafting effective prompts. The course is structured in a logical sequence, building your skills step by step. We'll start with an introduction to chat EBT and move into prompt engineering fundamentals. Later we'll explore advanced techniques, real world applications, and expert level strategies. Now, let's begin our journey. 2. Understanding ChatGPT: Let's begin by diving into how Chat GPT actually works. Chat GPT is a conversational AI designed to respond in a way that feels natural and engaging. It doesn't think like a human, but it generates responses based on patterns from massive amounts of data. The better your prompt, the better the response you're going to get. HatiPT works by analyzing texts and predicting the most likely next word based on what has already been written. It doesn't truly understand meaning like humans do, but it recognizes patterns and structures to generate coherent responses. The more context you provide, the better DAI can predict what you need. Chat GPT is an incredible tool for generating ideas, summarizing content and answering questions, but it's not perfect. It can be wrong, updated, or even confidently incorrect. That's why it's important to verify information and fine tune your prompts to get the best results. Chat GPT is only as good as the prompts you give it. A vague question leads to a vague response, while a well structured prompt can generate something truly useful. That's why prompt engineering is such an essential skill. It helps you shape the AI's output in a way that works for you. Let's take a look at HGPT in action. We'll explore the interface, adjust a few key settings, and run a sample prompt to see how it responds. This demo will give you a better sense of how HTGPT works and how small adjustments can impact the output. So to get started, you simply just need a browser. You can use the favorite browser of your choice. Here I'm using Chrome. Of course, you can use any other browser that you like. And all you have to do is navigate to chatbt.com. And really, the setup is really simple. You just need an email address, and again, you can sign up for a free account using your email and a password that you can create. And once you do that, you can simply login, and you will see exactly the same interface that I'm seeing right now. I'm currently logged into my account in chagbt.com. As you can see, OpenAI has done a fantastic job in developing a very simple and effective user phase. There's really not much to it here. Let's quickly take a look here on the left hand side. You got a few options, and this is the left hand navigation bar, and you can simply expand or collapse by going through this icon, and you can see the tool tip that says close side bar. You can simply click that to expand and collapse this depending on what you need. Um, if you don't need access to your previous chats, then you can just simply close it, so there's less distraction. Your chats will show up here in a historical sense as you conversate with Chat GBT, and you start new conversations and chats, and then you'll be able to access them here. There are a couple of features here, so you can start a new chat by clicking this button, you can search existing chats. Can go to the library. You have access to a service called SOR, which is a video generation tool for AI, and currently this is only available to paid subscribers, which are the plus and Pro. Not available for the free account at this time, and then you have access to different types of GPTs. On the center here, we simply have the prompt. So this is where it says, What can I help with, and this is exactly where you can put in your prompts. Now, you can type your prompts in this box or you can simply enable the microphone and you can just talk to it and it will convert the speech to text. And you can also use the voice mode, and this is where hatGPT can talk to you just like talking to a normal human being. Really depends on your preference. You can enable voice mode if you like talking instead of typing. And over here, you got a plus button. So if you click on this, you can upload files and photos for a variety of use cases. For example, if you have a file with some datasets that you want HA GBT to analyze, you can do that Excel file, PDF file, whatever, or you can simply upload photos and ask HAGPT to do certain things like edit the photos or use that as a reference and create a new photo and so on. And then if you click on Tools, here you have a bunch of different features. So depending on what you're trying to accomplish and use JAGPT for, you can select these options. So this is for creating an image which is currently using Dali on the back end. You can search the web. So when you're giving it a prompt, you can have HAGPT actually search the web for the most up to date information given the topic of your choice. You can write or code. So this is using logic. You can run a deep research, which is a really cool feature. If you want to research something, this will use both sort of searching the Internet and logic. To put together a well prepared research for you. And think for longer, this is simply this enables the reasonable model, and those are models such as 01 or oh three from OpenAI, and they're really good for solving logical problems, math, coding related issues, troubleshooting and things like that. So I highly recommend enabling this if you're dealing with coding, for example, or software development. And over here, if you click here, you'll see that we are currently using the free version or the free tier of HAGPTO course, if you click on Upgrade, I'll take you to this page and you can choose from various options depending on your need. However, for the purpose of this course, I chose to use the free tier to show you that there's so much you can do with HAGPT just on the free tier. But depending on your use case and need, the plus is a very popular option. I personally do have the plus subscription, and it gives you access to always gives you access to newer models, and also there's less restrictions. So for example, with the free tier, you can only generate three or four images per day. You can only use the research five times a month. So there's various different restrictions that limit your ability to use some of the features, but with plus, you don't have those. And you also get access to the service that is called SRA, which is a really cool video generation tool from ONAI. One last thing, you can also try to customize HAGBT to tailor it more to your sort of needs and tone. So the way you can do that is you can simply just click the profile icon and click on customize HAPT. Here, there's several fields that you can fill. So, for example, what should HAGPT call you? Here you can put in your name, and then HAGPT will try to personalize the responses, and it just makes it more sort of like a natural feeling of talking to a human being, like, you're talking to someone else. So it feels more personal, which is quite nice. What do you do? Here, you can put in your sort of like the job title, and this helps ChahBT to kind of tailor responses specifically to your job title. So you can put in project manager, project manager, software engineer, nurse, teacher, and so on, HR professional, whatever it is that you do. Here, this is an important one. What trait should ha GPT have? And if you hover over this information icon, you can see that this is really helpful in order to set the tone. So you can tell ha GIPT to set the tone to something like formal or professional. It can be chatty and casual or friendly. It could be opinated. You know, if you have questions with multiple answers, you can try to give your best one. And here you see some quick responses that you can add like chatty, witty, straight shooting, skeptical, traditional, and so on. And then here you can put in anything else in terms of your interests, values, preferences. And then, you know, you can say, I like hiking, I like jazz. I'm a vegetarian. So whenever you're conversating with Cha GBT about different topics, you will try to use these settings and traits to customize and tailor the responses to basically the settings that you have customized. And over here, you have enabled for new chat, so this will take effect for any new chat that you start. For now, I'm just going to leave everything empty and exit data this. And one last thing I wanted to mention is here you got this option called the temporary chat. So if you turn this on, this is actually, it tells you exactly what temporary chat is, so it doesn't remain in your history. So it says temporary chats won't appear in your history. And for safety purposes, they may keep a copy for up to 30 days, but after that, it gets deleted. So it's temporary they are not going to use temporary chats to train their models on, and then also memory is going to be off, so it's not going to be remembering things as you're giving it prompt. So continue, and you can see the UI is a little bit different. It's a darker theme, and it says, This chat won't appear in your history. So when you are having a temporary chat, then it will not show up on the left side bar here as part of your history. And that's pretty much it. Now we're ready to actually start putting in some prompts in chat GPT and see how it behaves. All right now, let's go through a couple of example scenarios and get a sense for what a vague prompt looks like versus what a detailed prom looks like. So let's go ahead and start with the following prompt here. And I'm simply going to copy paste that in, and it says, tell me about space, and can either click this button here, the arrow upward looking button or simply just click Enter on your keyboard, and this should get HachiPT going and start processing your prompt, and it's going to interpret and give you the output. Now, as you can see here, hATGPT started to process and give you the results. Space travel refers to the act of traveling beyond Earth's atmosphere. It's giving you a brief history with a breakdown, types of different space travels and why it matters and challenges and future space travels. Future of space travel. Now, as you can see here, this prompt itself is it's vague, right? Tell me about space travel. So it's kind of vague. It's not specific or focused. And because of this, you're going to the AI response is going to be very broad, and you're going to get a generic answer, which is what we saw. So now let's go ahead and refine the request for a more useful output. And in order to do that, I'm going to really make my prompt more detailed and specific. So let's go ahead and do a follow up prompt and this one says, explain the challenges of long duration space travel, focusing on radiation exposure and muscle atrophy. So you can see now this is a lot more specific and a lot more detailed focusing on just a couple of specific things here. So let's go ahead and run this. And now you can see that Chat GPT is giving us the output, and it's really not really talking about space travel in a generic sense, but it's talking about some of the specific focus and topics that we asked to sort of tell us about. So radiation exposure here talks about what it is, why it's a problem, and that's how you would midgate those type of problems. And then it's doing the same thing. With muscle atrophy. So again, it's not talking about space travel in a generic sense. It's really focusing the output on these specific areas. And you can see, again, sort of same thing what it is, why it's a problem and the mitigation strategies for that. And then at the end, it's giving you a really nice summary table. And also at the end, something that Chachi Vida started doing is it allows you to provide some ideas on how to further carry out the conversation. Doesn't mean that you have to, but it gives you a good idea. So for example, it says, at the very end, it says, Would you like an illustration or a diagram showing you how spacecraft mitigate these effects. So essentially, it's trying to provide some pre loaded ideas in terms of something that you may have thought of already and wanted to explore or something that you may not have thought. So from an ideation perspective, it's very helpful. Now, again, it doesn't mean that you have to continue with this, so you don't have to say yes and continue with this conversation. You can put in your the next prompt, whatever it may be. Now, as you can see here, given this more refined prompt, you can see that the AI response is a lot more specific and it's giving you a well researched answer. For our next scenario, let's go ahead and start a new chat. And what I'm going to do is I'm going to show you the access to the Internet functionality here. So, for example, let's say, I'm going to use the following prompt that says, tell me the latest stock market trends for 2024. So in order for HAGB to be able to answer that, it also requires access to the Internet. Now, the models over the past few months have become smart to a point where they know when they should access their Internet themselves. So these functionality were not previously available prior up to GPT four, but now it is. So you can just leave this as is and run the prompt, or if you want to, you can actually click Search the web, and then this allows you to for chat chi if you actually reach out to the Internet, grab the latest information, and then give you the results. So first, it goes to the Internet, it gets the necessary information, then it analyzes it, it interprets it, and then it looks for patterns, and then it gives you sort of puts the results together and then gives it to you as an output. Um, when it comes to the latest stock market trends for 2024. So let's go ahead and do that. And you can see here, you can see it says searching the web. And here, it's giving you sort of like a picture of the chart for what the S&P 500 looks like. And then here, it gives you a breakdown. So 2024 market recap and trends, talks about the broad based equity rally fueled by Tech NAI, record setting highs and widening leadership, volatility and policy, inflation rates, earnings strength, risk to monitor through 2025, outlook summary, and so on. So you can see that it was able to look through things. And then here, it gives you some of the available resources here. So on the bottom here, you can see it gives you some of the articles, like the top market trend news. Over here, it gives you the sources. So if you click on it, it opens the right hand panel, and this is where you get all this information and citations. So it's providing you all of these resources, which is great because if you click on any of these, then you can actually go and read that specific website or article where Chat GPT got the information from. And while we are here, I also wanted to point something out here, too. So there are several options here like copy. If you like this response, you can give it a thumbs up. If you didn't like it, you can give it a thumbs down. This enables the training of the model. So if it received a bad response, you click on this, then they know that, you know, when they're training their model that this response wasn't good, so they'll try to eliminate that um, you know, given all the datasets and information that they have. And by day, I'm referring to the OpenAI team and people who are training these models based on all the user prompts and specific information and training data. Now, this is an interesting one because if you have the free tier on the top, you cannot change the model through here. If you had the plus or Pro account, which are the paid subscription, you could choose your models from top here, but right now you can't. It's just by default set to the free tier model, which is 40 at this point, and that will change in the future, of course. But what you can do is change the models from here. So if you click on this arrow, you can see that you can set it to Auto. You can choose GPT 40, which is great for more tasks. You can do 04 mini, which is fast. You can choose 41, and you can try again. And again, these are some of the things that you can choose given the limited free tier edition. But you can choose this from here. And if you did choose that, you can see that there's a couple of arrows here. And when you choose this model, choose a different model compared to what it is right now. It's going to regenerate the response given your previous prompt and give you a slightly different output depending on how the new model is going to process your prompt. So a neat handy trick. If you're ever wanting to change, switch up your models and get a different response to the same prompt, then this is where you can do that. For our next demo, let's go ahead and try a multi turn conversation. So this is important for context retention. And what we're going to do is we're going to start with an initial query, and we'll use the following prompt. Again, this is an example prompt, but let's say, what are some beginner friendly programming languages? And here, if we run this prompt, AI is going to give us a list of languages like Python, JavaScript, you can see here, Python, JavaScript, Java, Scratch, Ruby, C Sharp, and so on. And this is a list. So this is exactly what we expected for the output from Chat GPT. But the thing I'm trying to demonstrate here in terms of context retention is that we can ask follow up questions without repeating the context. So you can do a follow up prompt without really, again, talking about you don't have to repeat the prompt again because in here, you're asking about really the context is what are some of the beginner friendly programming languages. So beginner friendly programming language being the focus of this prompt. Now, you don't have to repeat the same thing when you're asking a follow up question, such as which one is best for web development. So when ChaiPT is actually processing this prompt, it already has context from this same chat that you're currently in and also the previous prompt, it has the ability to remember. So when you're asking the questions such as this, which one is best for web development, it understands that you're referring to the beginner friendly programming languages. Let's go ahead and run this. And you can see here, it says for web development. The best programming languages are JavaScript, HTML and CSS, Python, PHV and so on. Again, it's able to retain context as you progress through the chats in Chat GPT. 3. What is Prompt Engineering: Now let's talk about what prompt engineering is. Prompt engineering is all about crafting inputs in a way that helps AI generate meaningful responses. The better your prompt, the better your results. It's like giving directions. Clearer instructions lead to better outcomes. Without the right prompt, AI can give answers that are too broad, inaccurate, or just not useful. A well crafted prompt helps you save time and get better results by making sure AI understands what you really need. The way you phrase a prompt directly impacts the quality of AI's response. A vague prompt leads to generic answers while a precise, well structured prompt delivers valuable insights. Let's compare the two examples, a broad request versus a focused one. A strong prompt includes four key elements, clarity, context, constraints, and examples. The more specific you are, the better chat GPT can tailor its response to your needs. For instance, adding a word limit or defining a writing style makes a huge difference. 4. ChatGPT Setup: Let's discuss setting up your AGBT account, and what are some of the tools you can use in conjunction with HGPT? To start using HGPT, you can go to OpenAI's website or simply go to chapt.com. The free version now includes access to GPT 40, which is OpenAI's latest model, as of the time of this recording. Of course, that'll change in the future. If you need more power, the paid plans provide extra features, faster processing and additional AI models and features. ChaGPT now offers multiple AI models. GPT four oh is the default and works well for most tasks. There are also smaller faster versions like GPT four oh, Mini and older models like GPT legacy. Additionally, OpenAI provides specialized models like 01 for advanced reasoning and 03 mini high for coding and logic. Open AI offers multiple pricing plans. The free plan gives access to GPT four oh, while the plus plan at $20 a month USD provides extended features. Businesses can opt in for the team plan at $25 per user or the P plan at $200 a month for high end AI capabilities. Choosing the right plan depends on how often and how deeply you use AI. Beyond the standard hat GPT experience, you can enhance AI functionality with plugins and browser extensions. These tools help automate workflows, streamline research, and boost content creation, making hATGPT even more powerful for work and learning purposes. To enhance your HATIPT experience, consider utilizing browser extensions like HGPT for Google, which brings in AI insights directly into your search results or Merlin AI offering on the fly assistance across various websites. For automating tasks, the ZapirPlugin connects HAPT to thousands of applications streamlining your workflow. If you're looking to boost computational capabilities, the Wall farm Alpha plugin is invaluable. In business settings, integrations like Microsoft copilot and V HAPT into everyday tools like Outlook and Excel, while the Canva plugin aids creating visual content effortlessly. These tools not only enhance productivity, but also expand the horizons of what you can achieve with JATGPT. 5. Practical Exercise: Now let's bring it all together. By going through a practical exercise here, I want you to go through your CHAT GBT account setup and run some basic prompts. This exercise will help you get hands on experience with HGBT. You'll create or log into your account, explore available models, and test different prompts to understand how HGPT generate responses. First, head over to HAGPT's website at either chat.copa.com or chatgpt.com. If you're new, sign up using your email, Google or Microsoft account. If you already have an account, just log in using the credentials. Depending on your needs, you can stick with the free plan or upgrade to a paid version for extra features. Once logged in, take a moment to explore the different models, just like we did in our demo. GPT four oh is the default and works well for more tasks, but there are also specialized models for reasoning and coding. Now, you may not have access to all models if you're just on the free tier plan. Adjust settings like response length and tone to fine tune your interactions. You can also customize ha EPT like we did during the demo. Now it's time to test HAT EPT. So start with a simple question, then refine it to be more specific. Experiment by changing the tone and format, ask for explanations in different styles to see how responses change, and this should help you understand how prompts shape AI's output. Think about what you've observed. Did adding more details improve the response? How did changing the tone affect the results? Understanding these differences is key to becoming an effective prompt engineer. The more you practice, the better you'll get at guiding CHAT EPT to give you the answers you need. 6. Well Structured Prompts: Let's now discuss the components of a well structured prompt. AI doesn't think like a human, I recognizes patterns in text. That's why a well structured prompt is crucial. The clearer and more specific your input, the better the AI's response. Adding structure and context ensures you get exactly what you need. A great prompt has four core elements, clarity, context, constraints, and examples. Being clear and direct helps AI understand what you're asking. Providing context gives necessary background while constraints refine the output. Finally, giving an example helps AAI match the format you want. Clarity is everything. A vague question like, tell me about cars could return anything from history to mechanics. Instead, a refined prompt like summarize the evolution of electric cars in under 100 words guides AI to provide exactly what you need. AI performs best when it has contexts. If you just say, write a product description, the response could be too generic. But by specifying that the product is for a children's smartwatch and emphasizing safety features, you get a more relevant, engaging response. Adding constraints help refine responses. If you ask, explain climate change, DAI might give you an overwhelming answer. But by setting award limit and specifying an audience, you control both the depth and the complexity. When you want a specific type of output, giving an example really helps HAT GPT. If you ask for a social media caption, but don't provide a reference, the AI may not match the tone or style you want. A small example can make a big difference. In this demo, we're going to break down a poorly written prompt and refine it, and we'll see how you can improve a weak prompt. So let's start with very generic and weak prompt. And as an example, I'm just going to say, tell me about space. So let's go ahead and run this prompt in hat GPT and see what it comes up with. Okay, so you can see here, it started giving us some answers. What is space made of talking about vacuum, stars, planets, moons, key features of space, now or sound, microgravity, Earth place in space, how we explore space, why space matters. So you can see here the response that AI has given us, yes, there is some useful information and it's informative. However, the AI response is too broad. It's covering history, explorations, planets, and so on. So now let's go ahead and refine it by adding specificity, and this is where prompt engineering really helps us get exactly what we need from AI. So instead of using a broad and generic prompt such as Tell me about space, let's go ahead and refine that for a stronger prompt. Now, I'm going to say summarize the history of space exploration, focusing on major milestones from 1950 to today. So I'm explaining what, right? I'm telling you what to focus on. So major milestones, and I'm giving it a timeline. So on this is no longer broad, it's no longer generic, it's focused and specific. So let's go ahead and run this. Now you can see the results of CHAT GPT is tailored to what we asked. So it's giving us the timelines 1950-1960, 1970s, 80s, 90s, and it's just telling us exactly what the progress has been in terms of the space exploration. And it's focusing on basically major milestones. So it's not really mentioning every single event is just focusing on the major ones, which is exactly what we asked for. And as you can see here, it is this response now is more structured. It's listing key events like the moon Landing and Mars Rovers. And also, you can see that it's a lot more like it's a lot more detailed given the requirements that we actually talked about. Now, let's go ahead and go take this one step further, and this is where I want to add constraints for tailored response. So now what I wanted to say is summarize the history of space exploration in under 100 words, highlighting three major achievements. So now I'm just restructuring the prompt and asking it to be more concise. So under 100 words, highlighting three major achievements. So I'm not really giving it any timelines from any specific year. I want you at GPT to actually give me the answer. So let's go ahead and run this. Now you can see it says since 1950s. And again, this is going back to the point in the previous lecture, which was context retention. Because in my previous prompt, I was talking about, you know, 1950s. I don't have to mention that again. I already it has the ability to remember that in its memory. So you can see it says it didn't just start from 1910 or 20 or 30 or 40. It started from 1950. So it's able to retain context from the previous chats in the same conversation. So since 1950, space exploration has advanced rapidly, and it's talking about 19:57, the Spot Nick one. And then it's talking about 1969, the Apollo 11 mission. And then in 2020, the SpaceX's new dragon, crew dragon. So three major milestones. I picked the three major ones for me, and now it's giving us the output. But if you look at this output here, you can see it's pretty concise. It's to the point, and it's relevant. 7. Types of Prompts: Now let's get into talking about types of prompts and their uses. Not all prompts work the same way. The way you phrase your question can completely change the AI's response. In this lecture, we'll explore the four major types of prompts, instructional, creative, exploratory, and conversational. Knowing when to use each one will make your interactions with AI much more effective. Instructional prompts tell AI exactly what you need. Whether you want a step by step guide, a structured summary or a formatted response, being direct helps AI deliver precise results. This is useful for productivity tasks like writing emails or summarizing information. Creative prompts are great for unlocking AI's storytelling and brainstorming potential. You can use them to generate fiction, craft compelling marketing copies, or even write poetry. When using a creative prompt, you can also specify tone, style, or perspective for even better results. Exploratory prompts help AI analyze topics in depth. Whether you're comparing ideas, looking for pros and cons or diving into industry research, these proms are excellent for structured insights. This makes them particularly useful for professionals and students looking to gather information efficiently. Conversational prompts make AI feel more interactive. Instead of asking one time questions, you can create an ongoing dialogue where AI remembers the context within the session. This is great for brainstorming, customer support, simulations, and personal assistance. In this demo, we're going to go through some live examples for different prompt types and we'll see prompts in action. So we'll walk through four different prompt times and show how AI responds differently based on the structure and intent of the prompt. So you'll see the original prompt. We'll see the AI's response. And we'll go from there. So let's start with instructional prompt where we provide clear structured guidance. So here we'll show how instructional prompts gives you direct structured responses and how AI follows specific instructions. So let's start with simple prompt. And for this example, we'll use the following prompt that says, summarize the key features of electric cars. So let's go ahead and run this and look at the results. Okay, so here you can see it says key features of electrical results or EVs include, and then there's a list of numbered list of responses here, so powered by electricity, stores energy, used to plug into external power source and so on. Now, you can see that this is sort of like a general summary of electric cars. So the AI response is basically just giving you a general summary. Now, let's refine the prompt by adding some structure. So for this follow up prom, I'm going to say the following. I'm going to say summarize the key features of electric cars in three bullet points using simple language. So this is more refined, and again, I am refining the prompt by adding more structure to it. So let's go ahead and run this. And here, you can see that their output is more concise and is following exactly our instruction, which was give me three bullet points. So here you got your three bullet points, and over here, it's using very simple language. So no gas needed, no exhaust fumes or pollutions, less maintenance and lower fuel costs, which is again, really simple to understand. So here you can see this set of results or output. This is when it comes to the AI response, this is concise, well structured list of key features. And here, when you're adding bullet points, it makes the response more structured and specifying simple language in your prompt ensures clarity. Okay, now let's take a look at creative From. So this is generating unique or imaginative content. So here you'll see how a creative From influences AI's tone and style, and we can highlight how AI can generate humor, storytelling, or even engage in content. So let's start by doing the following walk through. And here, we'll start with a very basic creative request. So I'm going to use the following example prompt. So it says, write a social media post about Mondays. So let's go ahead and run this. Okay, so as you can see the results, it's added some Imogs and then it says Monday mood, new week, new goals, same coffee, addiction, so there's some humor there. And you can see that again, it's a little bit sort of like it is funny. But again, we haven't really made anything we haven't really specified anything unique or specific here. So, DAR response is still somewhat general post about Monday. And what we can do is we can add, we can be more specific with the tone, and we can add sort of like a funnier tone to this. So in order to refine that for a funnier tone, I'm going to use the following follow up prompt. So here, now in my prompt, I'm going to say write a funny social media post about Mondays using a relatable meme style caption. So let's go ahead down and run this prompt. And here you can see that it says funny Monday me meme style caption. Me on Sunday night, I'm going to sleep early and wake up refreshed. Also me on Monday mornings wakes up with 17 alarms later wondering what year it is. So again, and some hash tags, of course, because we're asking for social media posts. And some Imoges, of course. And again, this is funny. And the AI response here is basically, again, you're changing the tone. You're being specific, and this is good because AI adapts when you define the tone and style, such as including meme style caption, which ensures AI aligns with social media trends. All right. Next, let's go through some exploratory prompt breaking down a topic for insight. So here, we're going to show how AI analyzes compares or explores different ideas, and we can demonstrate how structure prompts improve depth and clarity. So let's start by doing a walk through. And here I'm going to start with a vague research question. So I'm going to use the following example prompt and say, what are some productivity tips? So let's go ahead and run this. And you can see Chad GPT is giving us some output here, and it's categorized with mindset and planning tools and techniques, distraction control and self care, and each of them have a number list of items that you can sort of refer to as guidelines. But again, if you look at the AI response, it's a little bit long. It's focused, quite generic. So if you wanted to refine this for better structure, what we can do is use a following prompt and say, list the top five productivity hacks for remote workers with a brief explanation for each. So now we're not just saying, Hey, give me some productivity tips. We're being very specific. So we're asking for top five, and we're saying it's for remote workers and then provide a very brief explanation for each. So let's go ahead and run this. And this is now we're getting our five. And as you can see here, create a dedicated workspace, and it tells you why it works, separates work from personal life, and so on. Stick to a start and end time, why it works because establishing work hours maintains structure and prevents burnout. So gives you the details as well. But again, it's brief, it's concise, it's easy to read, simple to understand. And, you can see that the AI response is a number list with a short, clear explanation. No, this works better because a number list makes it easier to read, right? Adding this part to the prompt here for remote workers, adding that tailors the advice to a specific audience or group, and then asking for here brief explanation, this simply prevents excessive detail. Lastly, let's take a look at the conversational prompt. So this is where we are engaging in a multi turn dialogue. And here we'll see how HAGBT remembers context in a conversation, and we can demonstrate how prompts can feel more interactive. So for this walk through, I'm going to use an example which is going to let's start with an open ended query. So for this example prompt, I'm going to use the following sentence. I need help choosing a laptop for graphic design. So let's go ahead and run this prompt. Okay. So as you can see here, it's actually breaking it down in different numbered list categories. So what kind of graphic card design do you want? What kind of graphic design do you do? So it's asking you a little bit more context, print design, this, this, which kind of question. So right now, before actually giving us an answer, it's asking us a bunch of questions before it can give us the right response. So it's asking us what kind we do and here is providing some information like three D modeling, motion graphics, things like that. W software to use the most Ab Photoshop, Illustrator in design, operating system, Windows, Mac, and so on, portability, and then the budget. So it's asking all of this before we can provide the necessary output to our prompt. So now, it's again, we're seeing some multiple options if we can do that. But what we can do is we can sort of continue the conversation where we are actually providing some more information or contact. So what I can do is use the following prompt and say, I have a budget of $1,500, right? And then I'm asking it, can you recommend the best option? So I'm giving it a criteria, and then based on that criteria, I'm asking it to recommend the best option. So let's go ahead and run this. And here you can see that now that it has that criteria, it's providing some responses. So MacBook er, 15 inch M three released in year 2024, it goes for this range, and it's giving you some of the technical specs, and it's telling you it's great for these things. Number two, DLXBSs around 1,500. These are these specs, and this is what the grade for. And then you got the Zeus and so on. So you can see that this prompt has definitely refined choices based on the budget. Now, let's add another follow up for more details. So this is where I wanted to demonstrate that it can retain context. So instead of repeating the whole thing again, like I want a laptop that's within $1,500 budget range, I'm going to have the following prompt that says, I also want good battery life. Which one should I choose? So then if you run this prompt, it's going to basically choose one of the three that it already provided you in the previous interaction with IGBT. So first asked for the first we asked for recommending a laptop. It it couldn't give us the answers right away. So it asks us some questions, so it has more information before you can give us some options. Then we said, Okay, we have $1,500, which is our budget, give us the best option, and it provided three. And now, when you're asking this follow up prompt, I also want a good battery life. Which one should I choose? When you're asking this question, which one should I choose, it already knows that you're referring to these three here. So essentially asking, out of the three that you previously recommended, which one would you recommend for the battery life or better battery life? And this is where it's really AI is really powerful because you don't have to repeat the whole thing over and over again every time you run a prompt, you can just continue the conversation in a natural sequence, and natural flow. So you can see here, when you asked that question, out of the three, it recommended MacBook Air 15. It's the best in class battery plus Excellence screen and strong performance. It's giving you the key specs, and it's saying why it's perfect for design and now battery life. And then what to watch out for and the verdict and so on. So the AI response, this definitely further refines the answer, and this matters because AI remembers previous responses within this session or chat. And conversational prompting is useful for, you know, things such as customer support, recommendations and interactive tasks. 8. Prompt Length: Now let's discuss prompt length and detail and how much is too much. AI responds based on the information you provide, but finding the right balance is really key here. If your prompt is too short, the response might be vague but if it's too long, AI might get confused or lose focus. In this lecture, we'll explore how to fine tune prompt length for the best results. When your prompt is too short, AI doesn't know what you're really looking for. Asking, tell me about leadership will get a broad answer. But if you specify the types of leadership and request examples, the response becomes much more relevant. A prompt that's too long may overwhelm AI leading to inconsistent or incomplete responses. Instead of overloading it with too many instructions, keep your request focused. Asking for a clear comparison with a word limit keeps the response precise and useful. The best proms provide enough context without overwhelming AI. Use precise wording and clearly define what you need. If a request is too complex, breaking it into multiple steps can help get better results. 9. Improving Output: Let's now talk about improving output with contexts and examples. AI doesn't think like humans. It processes text based on patterns. That's why context is crucial. Without it, AI may misinterpret your request or provide a generic answer. Adding relevant details makes responses clearer, more accurate and tailored to your needs. When AI lacks contact, its responses are often too broad. Asking for a generic product description may not give you what you need. But when you specify details like the product type and key features, you guide AI to produce a much better response. If you want AI to match a specific style, providing an example is the best way to guide it. Whether you're writing a social media post, email or product description, showing a reference helps AI understand the format and the tone that you want. In this live demo, we're going to be looking at testing different prompts with and without context. And what we'll demonstrate is show how adding context improves AI generated responses. And we'll be starting with sort of like a vague prompt. Then we'll add context to improve accuracy and then further refining the prompt for clarity and specificity. So let's start with scenario one, and in this case, we're going to do a vague prompt versus a contextual prompt. And what we'll show here is the lack of context leads to sort of a broad or unhelpful response, and will demonstrate how providing details makes the response more focused. So, for example, let's start with a vague prompt. So here, I'm going to say, give me a summary of a book. So let's go ahead and run this prompt. Now, here, Chat GPT is not able to answer us, and it's asking, tell me the title of the book that you'd like to summarize. Now, given that we weren't able to provide or I should say, given that AI wasn't able to provide an output, given our generic prompt, what we can do is we can refine our prompt for contacts, so it can just jump into the output and start giving us some results. So instead, I'm going to use another prompt and this one should have more context. So this prom says, summarize the book Atomic Habits by James Clear in 100 words, focusing on key takeaways. So now we are giving it context, and we're telling it which book and exactly how we want it to be summarized. So we're saying basically in 100 words or less. And here you can see that the AI response, we're getting a paragraph, and that is exactly what we sort of are looking for, which is the summary of the book and you can see that the AI's response is simply a concise, structured summary covering the book's main points. And this refined prompt works much better because AI now knows which book to summarize. The hundred word limit keeps the response short and useful and asking for key takeaways ensures AI focuses on actionable insights. Now in our next scenario, we're going to look at lack of context and creative writing. I want you to know that the AI models have improved significantly. So something that I wasn't able to if HAHIPT came out late 2022, at that point, it was not able to process a lot of these type of problems. But over time, the models have gotten a lot smarter. There is more feature sets in the application. So it's not struggling like it used to. So again, things just get better over time. As more data gets as the product gets more developed and more training data becomes available through usage and other means. So here, what we're going to do is a lack of the lack of context and creative writing, and we'll try and demonstrate how AI struggles with unclear instructions and creative tasks and how adding descriptive elements really improve that quality. So let's go ahead and for this example, I'm going to use the following prompt, which is a very broad request. So I'm going to say, write a short story. So let's go ahead and run that prompt, and then ChaGPT will just go ahead and write a short story. As you can see, the output here is being processed. So the name it's giving you the name, the last light, and then here's a very short story. Now, this is as you can see, the AI response is a random generic short story with no clear theme or style. So what we can do is add context to improve the output. And in order to do that, I'm going to use the follow up prompt, which says, write a successful or sorry, write suspense full. So we're giving it some genre here short story about a detective solving a mystery in a future six City with a surprise ending. So now this is a lot more specific, and now we are telling Cha GBD, we're giving it the theme. And here we should see a much better output. So you can see, again, the format is the same. So it starts with the title, and then you see the story, and this is exactly what we were sort of looking for. And you can see that the AI response is now focused on suspense, mystery, and a futuristic setting. And this prompt this refined prompt is better because AI now understands the story's theme and genre, and the specific setting, which in this case, futuristic city and the tone suspenseful shapes the response better. And asking for a surprise ending also ensures AI follows a structured plot. 10. Practical Exercise: Now let's bring all of our learning together and go through a practical exercise. And here is where I want you to rewrite a vague prompt into an effective one. In this exercise, you'll take vague prompts and transform them into clear structured ones. You'll see how adding clarity, context, and constraints improve AI responses. By the end, you'll have a better understanding of how small changes can lead to big improvements. A weak prompt leaves AI guessing. If you ask, tell me about fitness, you could get anything from exercise tips to nutrition advice and so on. So without focus, the response may not be useful. That's why refining prompts is essential. By specifying the focus, in this case, strength training, the target audience, so say women over 40 and response length, for example, 100 words or less, we guide AI toward a more useful answer. This results in a clear, structured and relevant response. Now it's time for you to practice. Take the vague prompt, tell me about technology, and make it more precise. Think about which part of technology, who the audience is and what constraints will refine the response. Then compare your improved version to our example. Once you go finish going through this practical exercise, I'd like you to take a few minutes to think about your learnings and reflect back to see what takeaways you took from this. So think about how your improved prompt changed the AI's response. Did adding clarity help? How did context refine the answer? Did constraints make it more structured? These are all the key principles of prompt engineering that will help you get the most out of AI. 11. COT Prompting: Let's turn our attention now to some advanced prompt engineering techniques, starting with chain of thought prompting. Chain of thought or short for COT prompting is a method where we guide AI through step by step reasoning instead of expecting an instant answer. This approach is especially useful for complex tasks like problem solving, logical reasoning, and detailed explanations. When AI is asked a direct question, it often skips reasoning and gives an answer. But if we request a step by step breakdown, AI follows a structured approach, improving accuracy and transparency. And by the way, this is one of the exact reasons why Open AI has introduced that feature I showcased earlier, which is think for longer, and that essentially triggers a reasoning model. Chain of thought prompting is not just for math. It's valuable for decision making, troubleshooting, and even strategy building. By guiding AI through structured reasoning, we get clear more logical responses. In this demo, we're going to be creating a multi step reasoning prompt in real time, and the demo will compare direct versus step by step reasoning prompts just to show how chain of thought prompting improves AI responses. So we'll go through three different scenarios, a math problem, ensuring AI shows its work, decision making, which sort of demonstrates the structuring, logical comparisons, and problem solving for things like debugging code with, step by step analysis. So let's start with scenario one, and let's go through a math problem to ensure AI is able to show its work. So what we'll do is we'll show how a direct math question can only give you an answer, and then we'll demonstrate how chain of thought prompting ensures step by step explanation. So for this walk through, what I'm going to do is just start with a simple direct question. So I'm going to paste in the following prompt that says what is 32 times 47. Again, just a basic math problem. So let's run this. And there you go. You can see pretty straightforward 32 times 47 is 1054. So here, you can see that the AI response is 1,000 sorry, 1504. And the issue here is AI gives an answer, but no explanation. So let's go ahead and refine the prompt to request the step by step reasoning. So instead, what I'm going to do is use the following prom to achieve that. So I'm going to say, still the same problem, math problem. I'm going to say solve 32 times 47, but I'm adding step by step. So let's go ahead and run this. Okay, now you can see that we are seeing some sort of improvement. It's broken it down into steps. So step one, write the numbers, step two, multiply 32 by seven, multiply 32 by 40, add the two results, and the final answer is 1504. So here, you can see that you do see the improvement because now AI shows the complete breakdown of the calculation. And the key takeaway here is that the COT prompting helps AI show its reasoning, which makes the solution clearer and easier to verify. For our next scenario, we're going to be looking at decision making and structuring logical comparison. So here we're going to show how a broad comparison prompt can basically give you a very basic answer and demonstrate how COT structuring can organize the response more logically. So let's go through the following example, and I'm just going to start with a very simple question, and I'm going to use the following prompt for this, which says, which is better, remote work or office work. So let's go ahead and run this. And here you can see that it sort of DAI is giving a response, where it's saying that both have pros and cons, and then it starts talking about the pros and cons for each for remote work, and then the office work. So here, you do get a breakdown of pros and cons for each, but the issue is it basically lacks depth and there's no structured reasoning. So what we can do now is we can refine this with chain of thought prompting. So the way I'm going to do that is I'm going to use the following prompt here. Which says, Compare remote work and office work step by step, listing the pros and cons of each followed by a final recommendation. So let's go ahead and run this. Now you can see it's breaking it into different categories. So we got time management and flexibility, and then it's showing the pros and cons for each category. Then commute and location, communication and collaboration, productivity, mental health and social life, career development, visibility, and so on. And then at the end here, it's kind of showing you the final recommendation, and the best overall is actually a hybrid model, which, you know, you can go to the office certain days of the week and you can work from home, certain days of the week. So that's actually not a bad sort of like final final recommendation, assuming that your workplace allows that. So here, you could see the breakdown, and this is an improvement because the response is now logically structured with a well organized breakdown. And this sort of the key takeaway from this demo is that COT helps AI structure complex answers for easier understanding and decision making. Okay, so now for our third scenario, let's take a look at problem solving. So let's take a look at a debugging example where we debug code with a step by step analysis. And here, we're going to show how AI can identify and fix problems more effectively with structured reasoning. So for this example, I'm going to start with a direct debugging request. So for this, I'll use the following prompt, which says, fix this Python code for me. So first, I'm going to paste in the code, and then I'm just going to push this down a little bit and then say fix this Python code for me. And then here we have the Python code, and you can see that this is sort of the code itself, if you're familiar with this, but basically the problem here is that the code will cause a division by zero error. But let's go ahead and run this and see what ChaGPT comes up with. Okay. So here you can see that the Chachi PT has actually caught the err and says the code you provided will raise a zero division err when B is zero, and it says you can fix it by adding error handling using a tri acept block, and here's the corrected version, and then it will print this error. Okay, so this is fine. This is a fix to this potential issue in the code. And the issue here is that the AI gives you a solution, but it doesn't explain why. It made the changes that it did. So now let's go ahead and refine this with chain of thought prompting. So what I'm going to do is I'm going to use the following prompt in order to accomplish that, which says, analyze this Python function step by step, then explain potential errors, and then suggest a solution. So then I can just go ahead and put this prompt again, similar structure to my previous prompt. So put in this prompt or instruction because they're all part of the prompt. So I'm going to put in the instruction first, and then I'm going to create a new line here, and then I'm going to paste in the exact same code that we did up here. So just a different set of instructions in the prompt here, which follows the COT guideline. So let's go ahead and run this. Okay, now you can see hat GPT. Remember our prompt analyze this Python step by step. So you can see that it is analyzing the code, and now it's giving us a step by step breakdown. So function definition, it explains what it's doing, division operation, and explain what it's doing, function call, and then the potential error, which is division by zero. And it says in Python dividing by zero, raises this error, and then here's a suggested solution that it is sort of u providing to you and then there's some optional stuff here. And then some example outputs in terms of if you were to follow this fix as a guideline, here's some example outputs that you could potentially see. Now, here you can see the improvement compared to the previous response because AI now explains the issue step by step before providing the fix instead of just jumping into the fix. So the key takeaway here is that the COT prompting helps AI debug code in a structured way, making issues and solutions more clear. 12. Few Shot and Zero Shot Prompting: Now let's take a look at another technique called few shot and zero shot prompting. Few shot and zero shot prompting are techniques for guiding AI responses. In zero shot prompting, AI generates an answer without prior context relying on general knowledge. In few shot prompting, we provide examples first so AI understands the format or style we want. With zero Shell prompting, AI tries to understand what you need based on general patterns. While it can generate reasonable responses, it may not match your preferred tone, structure, or style without extra guidance. Few shell prompting gives AI a clear reference before generating responses. By providing one or more examples, you guide the AI to match a specific style, structure, or tone. This is especially useful in content writing, summarization, and complex reasoning. In this next demo, we're going to be comparing zero shot versus F Shot prompting for a real world task. And here, what we're going to demo is that we will compare Zero Shot and F Shot prompting by showing how providing examples improve AI generated responses. So the demo will consist of three real world scenarios. We're going to be writing an email, generating a product description, and summarizing an article. So let's start with Scenario one, and this is where we're going to write an email, and you'll see the zero shot versus few shot comparison. So we're going to show how a zero shot prompt gives a basic generic response, and then we'll show how a few shot prompt tailors AI response to a specific style. So for this example, what we're going to do is start with a zero shot prompt, and I'm going to use the following prompt for that. I'm going to say, write an email inviting colleagues to a team building event. So let's go ahead and run this front. Okay, so as you can see here, Chat CHIPT did its best to write an email, and it is pretty decent. So it's got a subject line. And it says, Hi team. I'm excited to announce that we're organizing a team building event, and it's at some player orders. So placeholders so you can put in the date time, location, activities. You can replace that with whatever you have planned. And then here you can put the RSVB date and then your name and position as placeholders, which you can fill out. So it's not bad. It's all right. It's pretty generic. So as you can see that their AI response is generic. There's no personality, and the issue here is that the response kind of lacks engagement, personality, and clarity. Okay, now, let's go ahead and refine this with a few shot prompt. And the way we do that is we're going to use an example to guide EAI in terms of what we're looking for, say, for example, for tone, right? So this is what I'm going to use as a prompt. I'm going to say, remember, Chat GPT retains contacts, right? So I don't have to repeat what I wanted to do initially, which was writing an email inviting colleagues to a team. So that was the main purpose of the prompt or this conversation up to this point. So now I'm going to use the following f so prompt, which says, here is an example of an engaging email tone that I like. So you're being very specific about what tone you like, and you're providing an example, and you want HATGPT to basically analyze that and follow the same guidelines in terms of producing an output that is very similar to that tone. So let's go ahead and run this and see what JAGPT is able to do. Okay. So now if you look at this email, it seems a little bit more different because the tone has changed. So get ready to hack, Hey team. I hope you're all doing great. We're thrilled to announce the upcoming event. Some of those things are still the same, so still because again, given that this is a team event, these things still are applicable, like the daytime, location, and theme. And these are placeholders that you can fill out based on the event. Uh, but here, you can see some differences, right? It says, This is a great chance to unwind, to get to know each other and so on. Please R SVB by this date. Here you can see that that has changed a little bit. And it says, whether you're coding, designing, pitching or just bringing fresh ideas, this is your chance to collaborate, build something cool, maybe even win a prize or two. So again, and then there's going to be food, swag, stuff like that. And then this part is still the same RSV B by the state. So you can see that it is following the tone that you're providing here, like, come ready for exciting activities, great food and some friendly competition, RSP B on Wednesday. So it is now being guided to follow the tone that you like. And it has able to successfully rewrite that email. So here, definitely, we can see the improvements because AI adapts a more engaging tone and a structured format. And the key takeaway here is that the zero shot AI responses are generic, but few shot prompts help AI match a desired style or tone. For our next scenario, I was initially thinking of doing a product description, but I thought we could change it to something a little bit more interesting and do a social media post instead because social media is a big part of our lives these days that I think it would be better and more real life example to go through the idea of a few shot prompting and see its use case. Here we're going to be crafting an engaging LinkedIn post, which shows the impact of the examples. So we're going to be demoing how AI produces a generic linked in post with a zero shot prompt versus how providing a few example post guides AI to match tone, structure, and engagement tactics. So let's go through the walk through together. First, let's start with zero Shall prompt. And here, what I'm going to do is say the following. So in JAGPT, I'll put in the following prompt that reads, write a LinkedIn post announcing our company's new mentorship program. So let's go ahead and run this prompt. Okay, so here you can see that Chachi PT has done a pretty decent job in creating sort of like a generic Linked in post, so exciting news from, and then the placeholder for your company name. We are proud to announce the launch of the mentorship program, and then there's some description here why we're doing this with some bullet points, and then just talking about sort of the shoutouts. And then the hash tag, of course, because this is a social media and hashtags are quite common. So looking at the output here, there's really nothing wrong with this, but one of the issues that I can see is it just feels flat and it reads like a press release. And also, it lacks a personal voice, storytelling, or a strong CTA or also known as call to action. So this is where we can leverage FusshotPmpt. And in order to do that, I'm going to use a couple of examples in my next prompt, which is going to leverage the fus shot prompt technique. So what I'm going to say is the following. So I pasted this in, and it starts like, here are two examples of the style I like example one, and then I put in my like an example, Linked in post. So here we're talking about Heat work. I'm thrilled to share that I've just kicked off our future leaders mentorship program at some made up company name, paired with an amazing mentor, and so on some hashtags, some Imogs. And then example two, same thing again. I'm using that. There's some dates in here and so on and then so I'm providing two examples here with the style with the style and tone that I like. And now I'm asking This is the prom. So I'm asking now, write a linked in post announcing our company's new mentorship program in the same style. So this is where you're leveraging this prompting technique. So let's go ahead and run this prompt. Okay, now you can see Chad GPT was able to produce this, and it says, based on yours on the style of examples, here's the linked and pose for your use case. So again, there is Emojis involved, so big news from the team at your company. We just launched whatever name of the program, our brand new mentorship program and so on. I just a few weeks of piloting. We've already seen powerful pairings. And then, again, applications are now open. Here's the deadline placeholder. Let's build the feature. So empowering empowering words and sentences. And then, again, this is the hash tags that are used for this. So you can see that the AI response, it's more detailed and it's more engaging and it it matches our tone and the examples that we provided. So here, the key takeaway from this demo is that by showing AI concrete examples of tone, structure, and engagement techniques, Fuso prompting helps you craft social media copies or really any other use case that you have that resonates with your audience and drives action. In this last scenario, we're going to be summarizing an article and we'll be structuring AI saput with F Shot prompting. So here we'll show how zero Shot prompting produces a random summary format, and we'll show how F Sha prompting improves the structure of AI summary. So for this example, we're going to start with a zero shot prompt, and what I'm going to do is ask it to summarize an article. Before we get started, though, if you do see this message on the bottom here, this is pretty normal. So if you're on the free tier, after using hat CPT for some time, within the same day or within the same session, you may run into this limit, and this is because the GPT 40 model is actually it's one of the things that is provided for the paid subscribers on the plus and Pro plan. However, the free tier also gets a chance to try it out for a certain amount of proms and time throughout the day. So it does reset again after some time. So just just ignore it. It changes it to a different model, which is a mini model, and I'll show you in a second. So you can just press X and exit data in case you see it in the bottom. But yeah, you can just ignore it. If you're on the free tier, you can still continue to use hATPT and it shouldn't really affect you that much. All right, so for this one, I'm going to ask this to summarize the article, and what I'll do is I just have a generic article here, but let me go ahead and use the following prompt, and I'm simply just going to say summarize this article in three sentences. Okay. And then Colin and then I'm going to come over here, and this is just a random article I picked from Google, and it says what the feature of renewable energy looks like. So if you go to this address, you can look at the same article. It's from earth.org. So I'm just going to copy this, go back into HGIPT and I'm just going to copy this or paste this in Because instead of copy pasting the whole content, which is something I can do, I can just put this in because HGPT has the ability to search the web, if you remember. So search the web is the functionality that it it has. So I don't need to copy paste it, but you could if you wanted to. You could just go in here, copy all this content and then paste it back into HAGBT with the same prompt. So let's go ahead and run this. And you can see it says searching the web, so it is activating that feature, and it is actually able to have access to that article, and it's able to summarize it. So here you can see it did the summary, it's a paragraph, so the article from it discusses the rapid growth and feature projections of renewable energy and so on. So here you can see that the AI response is somewhat unstructured and it may lack key details. And the issue is, again, lack of structure and missing key takeaways, because if you wanted to focus on certain things and key takeaways, this is not the best way or the most readable way for that. So what we can do is we could actually use a few shot prompt to improve that. And get it to the format that we actually like so this is what we can do. I'm going to use the following prompt that says, summarize the article, and ha GPD knows what we mean by that, right? So we don't need to let me just take this out because it already has the context. So summarize the article using this format. And this is where I can actually give it some structure. So I'm telling it to use this structure. By the way, I don't think we really need the code anyways here. So main idea a brief statement of the article's focus, key takeaways, three important points, conclusion, a one sentence summary, and now summarize the feature of actually, we don't even need this last part, either. Sorry, I'll just take this out because we are already telling you to summarize the article here. So let's go ahead and run this prompt here. Okay, so now you can see that it has a lot more structure. So you got the main idea. Remember, you gave it an example. So ChahPT is matching the response or its output with your structure and example. So it's following the same guideline you provided it. So you got the main idea. It explains the idea. It has three main key takeaways. So for someone who's busy who is not able to go and spend time and reading this whole thing, they just want to understand what are the top three key takeaways, they can just read this, which is a lot more brief and less time consuming. And then there's a conclusion here. So we asked them for a one sentence summary of the input and it's talking about the impact here in one sentence. So you can see that the improvement is huge and AI now can follow a structured format, making the summary clearer and more valuable. And the key takeaway here is that few shot prompting improves the structure and clarity of AI generated summaries. 13. Multi Turn Prompting: Let's now talk about multi turn prompting for continuous conversations. Multi turn prompting allows AI to maintain contacts across multiple interactions. Instead of starting fresh with each query, you can build on previous responses, making AI feel more like an interactive assistant rather than a one time answer generator. In a multi turn conversation, AI retains previous responses. Instead of repeating details, it continues the discussion naturally. This is especially helpful when refining ideas or exploring complex topics step by step. Multi term prompting is great for brainstorming, research and customer support. AI can refine, expand and improve responses with each interaction, making it feel more like an ongoing discussion than a single use tool. In our next demo, we're going to be keeping consistency and context in ongoing chats, and this is what we're going to demonstrate, which is going to be a demo that will showcase how AI maintains conversation contexts within a session and improves responses through multi turn interactions. So the demo will consist of three real world scenarios. We'll do a technical inquiry, we'll do some creative writing, and then we'll do some troubleshooting and problem solving. Then for each scenario, we'll be going through at the start with an initial prompt and then observe the AI's response. We'll follow up with the context dependent queries without repeating any other details, and then you'll see how AI refines and builds upon previous responses. So our first scenario, we're going to be doing a technical inquiry, and this is where you'll see AI remembers the topic of discussion. So you'll see how AI remembers what the user's asking about within the same session, and we'll demonstrate how multi term prompting allows in depth discussions without repeating any other details. So let's start with a generic question. So for this prompt, I'm going to use the following sentence that says, What is Python used for? So let's go ahead and process this prompt. Okay, so as you can see here, the response is good and it's informative. It's talking about what the programming language Python is used for. So data science and machine learning, web development, automation, and scripting, finance, game development, and so on. So the response is fine. However, we can observe that the AI provides a pretty broad overview of Python's use cases. So what we want to do now is we want to follow up with a specific question without repeating Python or the word Python here. So here, what we can do is we can use the following prompt and say which libraries are best for data science. And now question workrk and then we're not using the word Python. So let's go ahead and run this front. Okay, so you can see here, it says, great question. For data science, Python has a rich ecosystem of libraries. Note that I did not use the word Python here, but Chachi PT was able to retain that context from the previous one because in my previous one, I asked what is Python used for. In the follow up prompt, I only said which libraries are best for data science, and it's able basically retain that context and interpret exactly what I mean. So I'm not just asking data science in general across all programming languages, but without actually having to specify Python in my follow up prompt, it was able to understand that. So you can see it is saying Python has a rich ecosystem of libraries, and now it's listing those libraries for Python specifically. So Pandas, this one is the go to library for structured data like CSVs, Excel, you got NumPi, which is efficient array of matrix operations used by other libraries. You got data visualization, and here's a bunch machine learning and AI, data cleaning. So it's giving you all the libraries for Python. So again, this is the point here is that when you are following up with the specific question without repeating the word Python, AI remembers the user is really discussing Python and does not ask for clarification. So now let's go one step further and ask for a comparison of two libraries without restating the contact. So if you go back up here, you can see that over here, you see Pandas and you see Pandas and NN Pi, which is mentioned, this is the output by HachiPT, so Cha GIPT has produced a result. So what we can do is, again, we can ask for the comparison between these two libraries without repeating all the details. So I'm going to use the following prompt that says compare Pandas and NN Pi for handling large datasets. So let's go ahead and run this prompt. So here now it's creating a table, which is really a nice format because it's readable. It's easy to compare the two side by side. So it's talking about the different feature categories and then each of them, so you can see the difference pretty easily. And the final thing I want to mention is that one thing to observe is that AI continues the discussion smoothly without needing the redundant context along every step of the way. So the key takeaway here is that multi urn prompting eliminates repetition and enables deeper discussions. Now, let's go through an example where we do some creative writing, and this is, again, where we want to expand on AI's previous response. So here we'll show how AI can build on previous ideas for brainstorming or creative writing, and we'll demonstrate how ulturnPmpting refines and improves AI generated content. So for our first example, let's go ahead and start with the basic story idea prom. So I'm going to use the following prom that says, give me a short plot idea for a sci fi novel. So let's go ahead and run this prompt. Okay. So here's a plot idea. It's a paragraph, and you can see that hATPTs done a good job in delivering the simple basic story. And one observation is that AI does create a unique plot idea. So this is a good start. Now, what we want to do is expand on a specific aspect without restating the full premise. So this is where I can use a follow up prompt and say something like describe the main character. So here you'll see some characters mentioned in the plot. So let's see what Chat GPT is able to interpret and accomplish without us providing that redundant information. So all I'm saying is describe the main character. And over here, you can see without any further additional detail, Cha GBT was able to describe that character, doctor Amina Rao, which is exactly the one that was mentioned over here initially in the basic plot idea. So over here, you can see that it was able to pick that up without us really giving the character's name or anything. I understood. It was able to interpret the context, and this is a great improvement because AI remembers the original story idea and is able to build on top of it. Let's take this one step further and ask for more details on a specific plot twist. So in order to do that, I'm going to use the following prompt that reads, What is the major plot twist in this story. So let's go ahead and run this prompt. And then here, ChaGPT is giving us or creating us a plot twist with some details and, the only observation here that I'd like to share is that AI AI naturally builds on previous responses without needing the full backstory each time. So this saves you a lot of time from copy pasting and going back and forth over and over again. So this is a great feature. And the key takeaway here is that multi turn prompting helps AI develop ideas progressively making brainstorming more effective. Next, let's go through a scenario where we are troubleshooting and problem solving with a step by step debugging. So in here, we'll show how AI helps troubleshoot errors by following an iterative step by step process, and we'll demonstrate how multi turn prompting refines AI's problem solving capabilities. So for this walk through, let's go ahead and start with an error message and ask AI to diagnose the issue. So I'll use the following prompt that reads, I'm getting a division by zero error in Python. What's wrong? So let's go ahead and press Enter. Okay, so you can see that ChachiPT was able to detect the general issue. And actually, it even went one step further and provided a potential solution where it says how to fix it. And before a while back, it wasn't actually able to do this, but this shows how advanced the Cha chiPT models are getting and how quickly, over time, things are improving and the models are just getting more knowledgeable and more powerful with a long set of very feature set functionality and capabilities in Cha chiBD. So this is amazing to see because a most of the time you would actually have to ask specifically for the fix. So here you got the example cause of the error. So you got the explanation, and then here you get an example, and then here you get sort of like a one potential fix. There could be more, and then here you have the results. Now, let's pretend that it didn't actually give us the how to fix, or this is not the particular one that we're looking for in terms of solution. So what we could do is we can just without repeating the problem again, right, without mentioning that division by zero error in Python, let's just use a follow up prompt to ask AI to suggest a fix without repeating all the details. So for that, I'm going to do this. How can I prevent this error? So let's put that in. So here you can see that it is recommending that as an example code fix, it is recommending that we use an if statement, and there's a couple more. Use a try except block, use a default value or fault fallback, or then just sanitize the input to ensure this doesn't happen. So this is an improvement because AI is giving sort of like a very concrete an actionable solution, actually multiple solutions here, which could be useful depending on your use case and you have multiple solutions to choose from. So this is an improvement because it's an actionable solution. Now, what you can do is you can refine the solution for better usability. So you can use the following prompt and say, if you go to the bottom here, you can say, can you modify this to raise an exception instead? So let's go ahead and do that. And then this is back to the original one of the solutions that was try and accept, which was one of the solutions that JAG BT mentioned to us previously, which was number two here, it's actually putting that into use and it's showing us some code to see that in action. So over here, you can see that I didn't have to repeat myself regarding the context. So the observation here is that you can see it is actually putting the code for us, and then it is here's a usage, depending that if this was sort of our function, it's using that this is how you would actually call the divide function. You got the function and then you can call that function in the tri except block, and this would help prevent the error. So AI continues the discussion by refining its own previous response, and the key takeaway here is that the multi turn prompting makes AI an interactive problem solving tool, refining responses step by step. 14. Practical Exercise: All right, now let's bring everything together by going through the practical exercise, and this is where I would like you to apply advanced prompting techniques to solve a problem. In this exercise, you'll apply the advanced prompting techniques we've covered. So few shot, zero shot and multi turn and chain of thought prompting to tackle a real world problem. This hands on activity will help you see how structuring your prompts can improve AI generated responses. First, choose a task you want to complete using AI. It could be writing content, analyzing information, or solving a technical problem. For this example, we'll ask AI to help develop a launch strategy for an online course. Start by using zero shot prompting. Ask AI the question without any guidance. You'll likely get a generic response that lacks depth or structure. This shows why refining prompts is necessary for getting high quality answers. By using few shot prompting and providing an example first, AI now follows a structured format. This makes the response clear more organized and much easier to apply. Multi term prompting allows you to refine AI's response through follow up questions. Instead of one broad answer, you can iteratively improve and expand the details to get a fully developed solution. Chain of thought prompting ensures that AI thinks logically and sequentially. It helps structure answers step by step, making responses more methodical and insightful. 15. Productivity and Automation: For this next section, let's turn our attention to optimizing prompts for specific use cases, and we can start with Chat GBT for productivity and workflow automation. JAGPT is an excellent tool for automating repetitive tasks, such as drafting emails, summarizing reports, and generating structured documents. By integrating it into your workflow, you can work smarter, not harder, saving valuable time while maintaining high quality output. AI can handle a variety of workplace tasks from writing emails to summarizing reports. Instead of spending hours drafting responses or extracting key takeaways from long documents, let HAGPT do the heavy lifting for you. Whether you're writing an email from scratch or refining an existing one, hat GPT helps you maintain professionalism and save time. It ensures your message is clear, polite, and well structured without the effort of composing from scratch. In this next demo, we'll be generating a full work email thread using CHAIPT. So here, this demo will showcase how CHAIPT can automate email conversations from drafting an initial email to handling responses and refining messages. So we'll go through three different scenarios. One includes writing a professional email from scratch, the other one generating a reply based on a received email, and then lastly, we'll look at refining the tone of an existing email. So for scenario one, we're going to be looking at writing a professional email from scratch. And here, I would like to show how AI generates a well structured professional email, and you'll also see how providing context improves AI's response. So let's start with a basic prompt. I'm going to use the following prompt that says write an email requesting a project status update from a colleague. So let's go ahead and press Center. Okay, so the email is, as you can expect, it has the structure of an email, so you got your subject line requests for project status update. And then there's some placeholders that you can fill out. So the colleague's name, the project name, your name, and so on. So here, again, this email, it's not bad. It's asking your coworker for a project update. Now you can see that AI is able to. You see, if you look at the tone, I hope you're doing well. I wanted to check in on the status of the project. Could you please share an update, if possible? So thanks in advance, you can tell that the tone is polite, it's professional, and it's an email with a clear structure. Now, let's go ahead and refine the email to add some urgency without being too aggressive or sounding impolite. So what I'm going to do is use the follow up prompt that says, Make the email more urgent while remaining professional. So let's go ahead and run this prompt. So CHATPT is able to rewrite the email so you can see it's still maintaining the professionalism, so I hope you're doing well. I'm reaching out to request urgent update on the project. We are approaching a key deadline, so I need a clear picture of what the current status is. So there is a sense of urgency. So you can see, please send over a quick summary by a specific time and date. So we can address any blocker. So let me know if you need anything from my side, and thanks for your attention to this. So you can see the tone that's conveying a level of urgency while it's still maintaining professionalism throughout the email. So the key takeaway here is that AI can tailor email content based on urgency, tone, and recipient expectations. Now for our next scenario, we're going to be generating a reply based on the received email. So here we'll show how AI can generate a context aware email responds and demonstrate how AI adapts its response based on provided details. So let's go through walk through together. In this example, we're going to provide AI with an incoming email, so this is what I'm going to use as my prompt here. It says, Here is an email I received, generate a polite response. And now I'm just going to paste in an email, and of course, this is an example email. So I'm going to page that in here. You can see it's a subject project update request, and then hi, whatever your name is, thanks for reaching out, the project processing well, and so on. So let's go ahead and run this prompt. There you go. So HPT is producing the results here. So again, the subject is whatever it's going to be a re because we're replying to the email that was received to us. So you can see hi colleague name. And then this is where HGPT is being polite. So thanks for the update. I appreciate the transparency. I'm glad to hear the project is mostly on track, and I understand the technical issues that came up. Please keep me posted of any major changes to the timeline. So the observation here that we can all make is that AI generates a natural professional response based on the received email. Now let's go one step further and refine the response to request a more detailed update. So I'm going to use the follow up prompt that says, modify the email to ask for a detailed breakdown of what's left to complete. So here you can see it doesn't really discuss that. It says, if you have a brief summary of the issues or need a feel free to share. But now we are actually asking for a detailed breakdown of what's left using this prompt. So let's go ahead and run this. Okay, so again, HAT TBT is maintaining that level of professionalism in the email. So thanks for the update. I appreciate the transparency. So for the most part, this part is the same as the last email if you take a look here. But the change here is that it says, when you have a moment, could you please share a detailed breakdown of the remaining 20% of the work? And of course, you can change this to whatever that fits your use case. This is just an arbitrary number here. So it would be helpful to understand what's left to complete and the timeline is moving forward. Looking forward to your response. So over here, you can see that AI adds specificity while maintaining professionalism, and this is where you can really witness multi turn prompting, allows AI to generate responses that match the flow of an ongoing conversation. For third scenario, we're going to be refining the tone of an existing email. So here we'll show how AI modifies the tone of an email to match different communication styles. So first, let's go ahead and provide hat GPT with an existing email. For this, I'm going to use the following prompt. It says, Make this email more formal, and then I'll go ahead and provide something. Again, this is just made up for the purposes of this demo. Uh, so I'll paste in, you know, it's very short, just a sentence. So it says, Hey, place holder name. Just checking in on the project status. Let me know if you need anything, thanks. So it's quite short and somewhat casual. You can sense from the tone here. So we're asking HAGBT to make this more formal. So let's go ahead and press Enter. So now you can see the Chachi PT has actually asked for a more formal version of the email. So project subject this project status update, says, you're doing well. I'm reaching to check on the current status of the project. Please let me know if there's anything you need from myside to support progress. Thank you, and I look forward to your update. So this is a lot more professional and less casual compared to the previous email that we had. So we can observe that AI adjusts the language to be more polished and professional in this scenario. Alright, now, let's try a different thing and change the tone a little bit to casual, even though the first one is quite casual. Again, this is just something I made up for the sole purposes of this demo. But let's say our starting email was actually this. So what we can do or whatever email that you received in real life is sort of in this format, or this was your initial draft. And let's say we just want to modify the email to make it more casual and more friendly. So I'm going to use the follow up prompt that says, Now rewrite this email in a friendly and informal tone. So let's go ahead and run this. Here you go. So first thing you notice it's a lot shorter. It's just basically one sentence. It says, Hey, I just wanted to check in and see how things are going with the project. Let me know if there's anything you need from me, and thanks a bunch. Done. So quite concise, very short, very brief, friendly, less formal. And here you can see that based on this follow up prompt, AI is able to adapt the email to match the different communication styles. So the one takeaway from this demo is that AI can adjust tone and style to match different workplace communication needs. 16. Content Creation and Copy Writing: Let's now move on to HGPT for content creation and copywriting. HGPT is a powerful tool for content creation, whether you are writing blogs, scripts or social media posts. It helps generate ideas, structure content, and refine wording. It allows you to produce high quality writing in less time. AI helps streamline content creation across multiple platforms. Whether you need a structured blog, a social media post or a video script, HGPT can generate engaging and tailored content effortlessly. AI can help structure content logically, making it more engaging and reader friendly. Whether you need a strong introduction, well organized sections or a compelling call to action, HGPT ensures your writing flows smoothly. For this next demo, we're going to be creating a blog post from a structure prompt, and what we'll do is show how structure prompts lead to better AI generated blog content by refining the prompt in multiple steps. So the demo will consist of three key steps. So we'll be generating a basic blog post with broad prompt. So zero shot prompting we'll refine the prompt by adding structure and formatting. So this follows the few shot prompting technique, and then we'll be enhancing engagement by rewriting specific sections which basically is using the multiturn prompting. So let's go ahead. And for each of them, of course, we'll start with the simple prompt, just like we've been doing. We'll observe the response of the AI, then we'll modify the prompt to get more refined, and then we'll enhance the final content for better engagement and clarity. So let's start with our first scenario. So step one, we're going to be generating a basic blog post using zero shot prompting. So here you'll see how a vague or unstructured prompt results in a generic AI response. For this walk through, I'll just be using starting with a very simple request that uses the following prompt and says, write a blog post about work life balance. So let's go ahead and run this. Okay, so you can see CHAT EPT is still outputting the content of the blog here, but it is following the structure of a typical blog. So you got your title here, finding balance, how to make work life harmony a reality. Then you got just talking about some definitions. What if work life balance, white matters, five practical ways to improve. So here it's giving you five tips, the myth, and then the final thoughts here. So very typical and following a good structure of a typical blog post. Now the problem here is that the response is a bit too generic and lacks depth or somewhat of a structure. And the observation here is that the zero shell prompts often generate vague or generic content. Now moving on to step two, this is where we'll be refining the prom with structure. So this is where we are leveraging the fuhot prompting technique. And here you'll see how adding a structured outline to the prom results in a clear more engaging article, and it will follow our sort of formatting that we request here through using an example. So here, what we'll do is we're going to modify the prom to include some structure via an example. So what I'm going to do is I'm going to paste in the following prompt. Says, write a blog post about work life balance using this structure. So up to this point, it's still the same. This was the previous prompt here. But now we're telling you, still the same prompt, but we're adding using this structure, and now we're providing an example. So we're saying give me these four categories. So introduction, why worklife balance is important, common challenges, practical strategies, and then conclusion, key takeaways encouragement. Also at the end here, I'm just going to do a new line so it's easier to see we're going to tell it to use a friendly and engaging tone. If you wanted to use a different tone, then you can. So this is what we have for our prompt. Let's go ahead and press Enter and see what ChaGPT comes up with. Okay, it took a few seconds, but now Cha GPT has finished the output, and if you scroll all the way to the down, you can see that it's done processing the output. And if you take a look at this, this is a lot more it's got a lot more depth, and also it has the structure that we're looking for and also the information and content that we're looking for, because in the previous prompt, we didn't really say what content for CHA chiPD to talk about or to produce. So we are relying on HAGPT to figure that part out on its own. But on the second prompt, we're actually telling it exactly what the content for each category should be. So introduction, we wanted to talk about why work life balance is important. Common challenges, we want to talk about struggles professional face on a day to day basis with this type of work model. So the content is now tailored to exactly our needs and use case. So here we got our introduction, just like we had these four categories here, introduction, common challenges, practical strategies and conclusion. So that's exactly what we have introduction, why the balance is important, common challenges. We got practical strategies. This is where it's actually giving us some tips based on what can help with things such as time management, setting boundaries, health care, and things like that. And then we got the conclusion, and this is where we talked about the key takeaways and encouragement. So this is you can see, as you can observe, providing a structured prom can significantly improve the AI's output. Okay, now we have a pretty good baseline for our blog post here. So we're happy with what we have. This is the foundation for, let's say, a pretty good draft. Now, we want to be looking at enhancing engagement by rewriting sections, and this is where we are leveraging the multi turn prompting technique. So what we'll show in this next step here in our demo is how refining specific sections of AI's response make content more engaging. So first of all, let's go through an example, and here we're going to improve the introduction to make it more compelling. So here, if we go back, this was the introduction we had in section one. So we want to rewrite this to make it more compelling. So what I'm going to do is I'm going to paste in the following prom that says rewrite the introduction to grab attention and use a relatable example. And please note, again, we're using the multi turn prompt, so as you can see here, I'm not repeating any of the information because the contact is retained throughout the chat or this session. So let's go ahead and run this. And there you go. So now, Chachi PT was able to rewrite the introduction for it to be a lot more attention grabbing and have an actual example here. So this is, you can see, ever found yourself answering emails in your pajamas at 11:00 P.M. Wondering where the day went. So again, it's using an example and it's making it more attention grabbing. So now we can go ahead in our blog and just replace this one portion that has enhanced the introduction part of our blog here. And of course, as you can see, AI now uses a relatable statistic and a stronger hook. So now let's go ahead and make this strategy section more actionable here. So if we go back, we have the practical strategies, which was Section three in the original blog post here. So we'll go ahead and use a follow up prompt to basically make it more actionable. So in order to do that, I'm going to use the following prompt that says, expanded strategies section with real world examples. So let's go ahead and run this. Okay, so now it's basically rewriting Section three, but you can see that it's actually adding an example for each category. So we're talking about time management. You can see it added an example for the prioritize your top three tasks per day section. So, Maria, a marketing manager uses the top three method each morning, she lists three priorities on the sticky note and keeps it next to her laptop. Everything else waits until those are done, no matter how many emails pop up. So these are good examples. T blocking, it's the same one, and then you got the downtime, like a meeting. So if it's not on your calendar, it probably won't happen. So it's saying that, you can actually schedule that in your calendar. So that's for the first section, time management. It's doing the exact same thing for set boundaries, and it's doing the same thing for self care. And here you can see in terms of improvement that AI adds real world examples for better relatability. And this is where you can witness that multi term prompting allows AI to refine sections for better readability, engagement, and, of course, clarity. 17. Coding and Technical Queries: Let's now take a look at HPT for coding and technical queries. JTGPT is a powerful tool for developers, offering assistance in writing, debugging, and optimizing code. Whether you're generating you code, troubleshooting errors or improving performance, AI can speed up your workflow and enhance your problem solving abilities. JGPT can generate complete code snippets for various programming tasks. Whether you need a simple sorting function or a complex algorithm, AI can provide functional and efficient code. AI can quickly detect and fix common coding errors. In this example, hatGPT identified a division by zero and suggested a solution to handle the issue gracefully. This demo will show how HatchPT can identify debug, optimize and explain code using structured proms. So here, the demo will consist of three key steps. First, we'll be debugging a broken function and identifying and fixing an error. Next, we'll be taking a look at optimizing the code for efficiency and readability. And lastly, we'll take a look at explaining the changes made to improve understanding. So for this, we'll start with an issue and observe the initial response of HAPT. We'll use follow up prompts to refine the solution and then see how I can explain this reasoning. Now, for the first step, what we're going to demo is we'll show how AI detects and fix an error in a broken function. In order to accomplish this, what I'm going to do is I'm going to provide at GBT with a function that's containing a bug. So first, I'll start with my prompt, which says there is an error in this function. Can you find and fix it? And then I'll create a new line here, and then I'm going to copy paste the following code here. So now, just to quickly in case you're unfamiliar with coding, this is a very simple Python function, very basic. But if you don't understand this, don't worry. I can walk you through this, but this is simply just a function here that is called calculate total price. That's the function name, and this is the definition of that function in terms of what it does. And it simply adds the price and then the tax, and then it returns the total. Now, one thing this function is now in this prompt is being passed in with a bug, and that's intentional because we're purposely including a bug JGBD can catch this and fix her. Now, the bug is in the function definition, we have two arguments. So we can pass in two arguments. One is the price and one is the tax rate. So when the function is being called, we have to pass those two things in so that the function can do its job. Now, over here, this is where we actually calling the function, which is inside a print statement, but not that we only have one argument, and we're passing in the first argument, which here is the price. So we're passing in 100 for the price, but we're not passing the second argument, which is the tax rate. So let's go ahead and run this and see what HAGBT comes up with. And there you have it. So right now, it says, Yes, there's an error with this function. The function, calculate total price expects two arguments, price and tax rate, but you're only passing in one argument when calling the function. So that's exactly correct, and that is the issue with this function. So it says this will raise a typer because tax rate is missing. It's giving us a fixed version. And here, it says it's actually providing us a couple of solutions, not just one, which is nice. So it says provide both argument. So here, for example, if you have a 5% tax, you would call the function. Well, first of all, you have to include the argument, and then as part of that, you're passing in a number, which in this case, is a tax rate. So for example, here, we're calling the function, we're passing in 100 for the price, and then we're passing 5% for the tax rate. And now it's able to do its job, make the calculation, which is price plus the tax, and then it will return the total. Option two, it says provide a default value for the tax rate. And the way you can do that is you could do that in the function definition. So you can see that this is how you would do that by providing value. So argument is called tax rate, but then there's a equal 0.05. So this is called the default argument. So if in a scenario like previously in our prompt, somebody does not provide the tax rate, it will use that default. So it becomes an optional and then argument and then it will actually use this one, and over here. So over here, you can see it's now calling so inside the function definition, we provide the default argument and we assign a 0.05. So that'll be the default value for that argument. And then when we're calling it, they're calling it using two different ways. So the first one, they're just passing the price, but this way, the function will still not run into errors and will still execute and return the result. And the way it does that is it's using the default value of 5%. So you pass in the price and it will return the total of the price plus 5% tax. Then the second one, you can pass in the actual argument, and please note that whatever argument that you're going to provide is going to supersede whatever the default value is. So the value that you're providing as a user into that argument for that argument in that function will always take precedence over the default value. The default value is there if you do not provide a value or provide a value that is different than the default value. So this is basically the fixed version. You can use whichever you like in terms of the use case or what you see fits your use case the best. But the point here is that HGBT was able to apply a fix and AI was identified was able to identify the function was missing a required argument in the function call and added the validation check as well. So over here, actually, that's one thing that we could have done. We could have thrown a if statement. AI did not come up with that, but if you wanted to, we could actually ask it to do that. So we could actually go ahead and ask it to add in sort of like a ifelse statement to add validation. So here you can see AI does help catch missing parameters, syntax issues, and logical errors quickly. On that note, let's go ahead and actually do a follow up prompt to add a validation check. So this is where we can actually have a logical check that can catch the error or return an error with a message in case, you know, the second argument in this case, the tax rate is missing. And it should be very easy to do that. Sometimes really depending on sometimes JGBD does include the if statement in here. In this case, it didn't you got the definition here, and then over here, you got another solution where it gives you the default value, which is fine. But we didn't get one solution where we have validation. So let's go ahead and do that. So Um, all you have to do is really say, add a validation check. So let's go ahead and do that. Okay. And there you go. So here you have your new function, and this one is adding validation check not just for the price, sorry, not just for the tax rate, but it's adding it for both the price and the tax rate. So here it says validate inputs. So before you process or actually going to calculating the total, it actually runs those validations to make sure everything's okay. So it says if your price, if you read this, it says, if the price must be an integer or a float and greater than zero, Um, actually, let me bring your attention down here as well, because it actually the nice thing about hATPDs actually explains the logic to you after it prints out the output. So if price the price has to be a float, right? Or it has to be integer or float, and the price has to be uh, graded or equal than zero. If not, it's going to raise error, and then it will throw that this error message price must be a non negative number. And then the same thing for tax. So here we have some validation. It says that the tax rate must be a number and it has to be graded than zero. So this could be whatever, like 0.05 as we saw in the earlier example or 0.1 for 10%, whatever. So over here, you can see that this added the validation check for those. And then once you validate both inputs, then you can actually go into your calculation, which is price plus tax and then returning the total. And again, here, you could actually because we have this default statement here as well, you can call the function a couple of different ways, and it'll still be valid. So one is just pass in the price and not the tax rate, which it will then use the default value of 5%, or you can just pass in the price and the tax rate. So this 10% here we overwrite this value here and it will do the calculations using 10%. Okay, now let's take a look to see if we can optimize the code for efficiency, and this is where I'll show how AI refactors the code for better readability and efficiency. So for this example, let's ask ChaGBT to actually optimize the function. So in order to do that, I'm going to use the following prompt that reads, optimize this function to make it cleaner and more efficient. So let's go ahead and run this. Okay, there you go. So here you can see you see this is a little bit lengthy now, right, especially with the added validation checks. So now it's actually reduced that it's more readable, it's cleaner, so you can see the improvements on the bottom here as well. So it says the types, float, clarify expected input and output types, compact validation using the N keyword there. And so this one here, and then it says math simplified, and then the default tax rate still applies, if not explicitly provided. So now, this is a cleaner, more readable function, and this shows that AI improves efficiency by reducing redundant code and making functions more robust. Now, the next thing I want to demo is you can always request Chat GBT to provide you an explanation of the fix or whatever it's doing. So sort of like it's thought process, explain its thought process to you or the reasoning or the logic that it's following. So here I want to show how AI can explain its debugging and optimization steps for better understanding. Now, in this scenario, it actually did that for us without us having to ask. So it created after our prompt, it gave us the improved code based on our requirements. But then there's a section called Improvement, so it actually explained everything. Sometimes it does that, sometimes it doesn't. Again, as the models get better and improved, it will be able to also from your previous chats and data and your interaction and based on its memory and some of the customizations you've made to CHAGBT, it'll be able to kind of follow and be able to predict what is it that you're looking for, depending on, for example, your job title or what you've asked it. So this is really important, but in this case, it did provide the explanation of the improvements. But let's get a little bit more technical. Let's say it didn't or perhaps this is not that in depth. Okay? Maybe it's just very high level. So what we can do is we can ask AI to actually explain the improvements made. So in order to do that, let's use another follow up prompt that simply says, explain the changes made to optimize this function. So let's go ahead and run this. Now it should be giving us a more detailed explanation. And again, the point here is we want to learn. We don't want Ja JB just do the work for us and move on. We want to actually be learning. And the way we learn from this is by following its thought process and understanding the logic it followed. So here you can see the first step, it says Tienes added, and now it's referring to the line that it changed. But then it also talks about the Y and the benefit. This is really important, especially it's very valuable when it comes to taking away learnings and learning lessons from this. Ask clarity about what types the function expects and returns. The benefit improves code readability and helps tools like inters and IDs provide better suggestions than error checking. So very logical, very helpful. Next one is the condensed validation using the NI key. It says combines the two separate validation checks into a single readable line. Less repetitive code, easier to maintain and expand. Great. And then simplify calculation. Here we have return price times one plus tax. It says replaces the more verbos one, and over here, which was this guy. So now instead of these two lines, we just have one line. So the benefit to this is shorter, cleaner, and mathematically equivalent, so you get the exact same thing. It doesn't return you a different response. And it also avoids the need for temporary variable, which is a very common classical problem, programming problem. And then improve readability and efficiency. So it says by combining all of the above, the function is now easier to read, slightly faster due to fewer operations and more robust to validation and type annotations. So this is how you can actually get AI to help explain the thought process in fixing or debugging the code, and this shows how Chachi BT can act as a mentor by explaining debugging and optimization steps in a clear and structured way. 18. Research and Learning: Let's now discover the use cases on HGPT for research and learning. So HGPT is a powerful tool for research and learning. Whether you're trying to summarize a complex topic, fact check information or synthesize insights, AI can accelerate the process and provide structured, easy to understand answers. AI helps researchers and students by condensing complex papers into key takeaways, making learning faster and more efficient, so you don't have to spend hours reading. This is particularly useful for summarizing academic research, policy reports, or any other dense material. AI helps fact checked statements by cross referencing information with multiple sources, allowing users to verify accuracy and detect potential misinformation or perhaps bias. Alright, so now let's do a walk through where we can use HAGBT to summarize, compare and fact check. And here we'll be summarizing a complex topic into key takeaways, comparing two perspectives on controversial issue and fact checking a claim and evaluating its validity. So let's start with step one where we summarize a complex topic. And here is what I want to show how AI AI breaks down a dense subject into clear digestible insights. For this, let's start with a very broad research topic. So I'm going to be using climate change. In order to do that, we're going to use the following prompt that says, summarize the main points of climate change research in five bullet points. So let's go ahead and run this. Okay, so you can see it was able to give us the five bullet point. The topic of climate change, I mean, it's so huge and so vast and there's so many things to go through, you know, hundreds of thousands of pages of research and articles. But if you are just trying to get sort of, like, break it down into a digestible insight, using that five bullet point technique is a great way to get started. Again, this is a start but the key takeaway here is that you can see ChatGBT was able to produce five bullet points on the topic, and this showcases that AI helps researchers and students quickly understand major themes of a topic. Next, let's take a look at comparing two perspectives on a controversial issue. So for example, or perhaps you might not find this controversial. It's a pretty debated topic nowadays. But here, we're going to show how AI presents arguments from both sides of the debate. So for this example, I'm just going to use remote work versus office work. So for this, we're going to ask Chat GPT to compare the different viewpoints. In order to do this, I'm just going to use the following prompt that says, compare the pros and cons of remote work versus office work. So let's go ahead and press on Inter. Okay, so Chat GPT is able to provide the answer here. And you can see it's able to so it did a pros and cons for each. So remote work, there's the pros, and then we got the cons here. And then for office work, we got the pros, and then we got the cons here. And then, of course, this is a really nice handy thing. It's provided the very very condensed and very readable, small summary of table that kind of just conveys the same information in just one or two words, which is really nice because it's very readable and it's easy to digest in very simple terms. Now, let's follow up to refine a specific aspect of this. So in order to do that, I'm going to say, using the following prompt, expand on the impact of remote work on productivity. So let's Enter and now it's going to sort of pick that apart a little bit and focus on that specific aspect of the topic. So positive impact on productivity. You got fewer interruptions, customized work environment, and of course, each of them have their more explanation and detail, flexible scheduling, less commuting time and autonomy. And then there's negative aspects. So communication delays, isolation, overwork and burnout, so all these things. And of course, a summary, overall takeaway. But the key here is that you can see HAGBT provides a balanced perspective, making it useful for research, debates and decision making. Okay, this next one is going to be interesting because we're going to be fact checking a claim for accuracy, and this is coming from the AI. But remember the data that AI is using, it's heavily trained it's heavily trained on datasets, and also it has access to Internet. ChachPT couple of years ago, did not have this functionality, but now it has access the features there. It can research and read all the articles online in order to fact check certain things. So here, we'll show how AI analyzes statements and verifies them against actual reliable sources, which is really cool. So now, let's go through this example, and we're going to ask AI to fact check your claim. And here I'm going to use drinking coffee. So I'm going to use the following prompt that says, fact check this claim. Drinking coffee dehydrates you. Of course, you can use anything. This is just for the purposes of this demo, but let's go ahead and see what it comes up with. Okay. So here, it's basically saying this claim is mostly false, but you can see that it says what the science. So talking about caffeine, the type, you know, the effect is mild and so on. Moderate coffee consumption, three to five. This doesn't lead to dehydration in healthy adults. And then because coffee contains mostly water, so it contributes to your daily fluid intake. And there's a supporting evidence. So it says at 2014 study in plus one found that coffee hydrates people cellularly to water when consumed in moderate amounts. So that's very interesting because a lot of people think it dehydrates you and according to ChaGBT does not. And then here's the bottom line. So it's giving you the facts, and then the summary here, drinking coffee does not dehydrate you if consumed in moderation. It's fine as part of your daily fluid intake, especially for regular coffee drinkers. However, excessive caffeine may have stronger, different types of effects, not necessarily dehydration. So this is interesting. So it is able to fact check this particular claim, but now it was able to produce the results, right? So we know what it's saying. And in this scenario, it's saying that coffee does not dehydrate you and this is false. But now let's actually follow this up by asking by a follow up prompt. And here, what we're trying to accomplish is we want to follow up by asking for source validation. And this is really important. So in order to do that, so we want to ask you, like what source are you using to come to this conclusion? So here, I'm going to use the following prompt that says, What sources support this claim. So let's go ahead run this. And then this is where it's actually starting to go through the sources and list them one by one and even giving you a link to that study. So this is all based on research and data, which is really nice. So the first one is the plus one, 2014 study. Here's a link, and it says talks about the title, the authors, and then the key finding in that study, and then you can read the whole article by clicking on this link. Next one Institute of Medicine in 2005. Same thing, finds the key finding that study, and then the link if you wanted to sort of the title, the source, and then the link, European Food Safety Authority, Mayo Clinic, and so on. So this is where AI can verify claims by analyzing multiple sources and providing evidence based responses. 19. Marketing and Sales: There are other use cases where HAGPT. So for instance, it can be used for marketing and sales. So let's turn our attention to these categories here. HGPT is a powerful tool for marketers and sales professionals. It can generate persuasive content, help craft targeted sales messages, and optimize customer engagement. Whether you're writing an ad, an email or a landing page, HGPT can make your messaging more effective and efficient. AI helps marketers generate multiple variations of content quickly, allowing for AB testing and optimization. Whether you're crafting social posts, ad copy or email campaigns, hATTPT makes content generation seamless. Great marketing copy focuses on benefits, not just features. AI can help refine your message by making it engaging, emotional, and action driven, encouraging higher conversations. Alright, now let's take a look at crafting and testing marketing prompts for social media. So here, we'll be generating multiple versions of a marketing message, adjusting tone and style for different audiences, and then creating effective calls to actions also short for CTAs. And we do that to drive conversations, and perhaps those could lead to convergence and sales. So let's start with step one, and this is where we'll be generating multiple versions of a marketing message. So here we'll show how AI creates variations of the same message for AB testing. So let's start with a basic product promotion. So in order to do that, I'm going to use the following prompt. I'm going to say, write a Facebook ad for a fitness app that helps users track workouts and diet. Okay, so now you can see HABT provided us with one variation. So here it says, ready to take control of your fitness journey, mid fit track. You're all in one fitness campaign and so on. So here, the features log workouts with these, track meals and macros, get real time progress, and so on. So it gave us one u variation of this Facebook ad for this Fitness app. But one thing I want to mention is there's a couple of things now you can do if you're into marketing and you want different variations, you can simply just ask HAGPTGive me different variations. But note here also note here that says, Would you like HAPT actually followed after um outputting this response here, it says, Would you like variations tailored for different audiences. Example, beginners, athletes, busy professionals. So this is interesting because it's predicting what you may need next, right? It's thinking that maybe you're a marketing professional, so you perhaps want more. In this case, you want different variations. So actually, let's just respond by saying yes because it's asking us, would you like versions tailored for different audiences. So let's say yes and see what it comes up with. So here, it's actually coming up with exactly what was in the bracket. So you can actually add more. It's just following with these, right? So it says for beginners, it gave us a variation. For athletes, it gave us another variation, for busy professionals, it another. And then it says, Want me to create versions focused on other features or tones, right? So the other things you could also do is you could also give it other tones or categories. So for instance, you can do motivational, you can do social proof. So you can actually ask for that. So let's actually go ahead and do that. Let's say, give me variations for, let's say, give me variations for motivational, data driven, and social proof. Let's run that prompt. And there you go. You get three more variations. It says motivational. This is the post or Facebook ad, data driven, and so on. So this is actually pretty good because now you can see AI generated variations allow marketers to test different messaging styles and optimize the engagement based on that. Now I want to show you how we can be adjusting tone and style for different audiences. So here we'll see how AI adapts the same message for different customer personas. So for this example, we'll just be continuing building on the previous example, but we're just going to be modifying the prompt to tailor messages to different demographics. So, for example, I'm going to use the following prompt to target three new demographics. So I have rewrite this ad for a fitness to appeal to three different audiences. So we got young professionals, busy parents, and senior citizens. Let's go ahead and run this. And now ChaGPT is going to be able to tailor these versions of the fitness app for each audience. So you can see now it was able to do that, young professional, busy parents, and for senior citizens. So here you can see AI can customize marketing messages to resonate with specific target audiences. Lastly, let's take a look at creating effective call to actions or CTAs to drive conversions. So here, what we can do is show how AI crafts persuasive call to actions for different platforms and marketing goals. So for this example, let's ask AI to generate multiple CTA options. So what I'm going to do is use the following prompt that says, generate five call to action or CTA phrases for a free trial sign up. So let's go ahead and press Enter. Now you get the call to action, and this is usually the thing you put at the end of your ad or website or whatever it is you're trying to wherever place your funnel, you're trying to drive sales. But here you can see that we're asking AI, and now it was able to come up with five call to actions. So start your free trial today. No credit card needed. Try it free for seven days and see the difference, unlock your fitness potential, join free for one week, and so on. So this is good. We can do a follow up prompt where we can make the CTAs more urgent. So in order to do that, like more time sensitive, for example, in order to do that, I'm going to use the follow up prompt that says, Make these CTAs more urgent and time sensitive. So let's go ahead and do that. And now we should really focus on the limited time offer type scenario. So start your trial now offer soon. Don't miss out, limited time only. Wells fill fast, act fast. So this is now giving more urgency to the call to action phrases over here. So here you can CAI can optimize CTAs to maximize conversions through urgency and action driven language. 20. Practical Exercise: Now let's bring everything together by going through a practical exercise where you get a chance to develop prompts for your own industry. This exercise will help you develop and refine chat GPT prompts tailored to your industry. Whether you work in marketing, finance, healthcare, technology or another sector, this hands on activity will allow you to create AI powered solutions that fit your specific needs. Think about one key task in your industry where CHAIPT can assist. Whether it's writing, analysis, sales outreach or customer service, this exercise will show you how to optimize AI generated responses for your workflow. Start by writing a basic prompt for your chosen task. Then analyze the response and see does it need more structure, clarity, or detail. Identifying these gaps will help refine the prom for better results. Refining prompts iteratively improves AI's responses. By adjusting phrasing, specifying tone, and requesting structured outputs, you can optimize AI generated content for your industry's needs. 21. Understanding ChatGPTs Limitations: Now let's talk about understanding HGBT's limitations. While HAGPT is a powerful tool, it has limitations. It does not think or understand like a human. It generates text based on patterns and data. This means AI can hallucinate, reflect biases or provide misleading information, making human oversight essential. One major issue with AI is hallucinations. It may generate false facts with absolute confidence. AI does not know facts the way humans do. It only predicts what text should come next based on its training data. AI is trained on large datasets that may contain biases which can lead to unfair or misleading outputs. It's important to review AI responses critically to ensure they are inclusive and fair. AI lacks common sense, reasoning, and deeper contextual understanding. When prompts are ambiguous, it may misinterpret the meaning requiring more specific phrasing for accurate responses. 22. Troubleshooting Poor Responses: Let's discuss how we can troubleshoot poor responses in chat APT. AI sometimes misinterprets prompts or provides vague answers when there isn't enough detail. By improving prompt structure, we can get more relevant, consistent and insightful responses. AI can provide unclear, inconsistent or overly long responses. The key to troubleshooting is adjusting prompts to get more structured, concise, or fact based answers. The more specific structured and guided your prompt, the better AI's response is going to be. Small changes in phrasing can dramatically improve the quality and relevance of the output. In this next demo, we're going to be debugging and refining food chat GPT responses. Here we'll demonstrate and we start off with a vague or problematic AI response, where we find the prompt step by step to improve clarity and accuracy, and then we'll use multi term prompting to guide AI to our better output. Now, for our first step, we're going to be fixing a vague or generic response. And here, I'll show how AI can produce vague answers when prompts lack detail. So for this next example, let's start with a generic request. And for this, I'm going to use the following prompt here. And I'm simply going to say, explain leadership. Okay, so as you can see, HAPT is starting to output the response here, and it's starting to explain the readership and starting with an introduction about what leadership is, it's getting into the key aspects of leadership, types of different leadership style, and then it has a conclusion paragraph at the end. So this is good and a good start. But you can see that the issue with this is that the response is too broad and it lacks real and specific insights. Now, let's go ahead and refine the prompt for a better answer. And what I'm going to do here is use the follow up prompt that simply says, explain three key qualities of effective leadership with real world examples. So here, we're going to be expecting more specific AI response. Here you can see that it's breaking it into different categories of leadership. So we got visionary thinking. It's giving you the definition, and it's giving you an example here, just like we asked in our prompt. You got empathy and emotional intelligence, you got decisiveness. So these are some of the key aspects of leadership, along with definitions and examples. And here, again, you get a short quick summary which is quite concise and readable. So what I'd like you to take away from this example is that adding structure and example makes AI responses more detailed and more useful, specifically for learning purposes. Now, for this next example, I would like to demonstrate how we can be fixing an incorrect or misleading response from Chat GBT. And here, we're going to try to show how AI sometimes provides factually incorrect responses. Now, use the word try because we can't predict if AI is going to give us the incorrect response or not. Most of the time, it'll give us the correct response, but from time to time, there could be misinformation, and that's why we should always be fact checking the AI output. So it's going to be very tricky to reproduce an error in a specific scenario. So we'll try and prompt and see how things go. So here, what I'll do is I'll ask AI a factual question. And for this example, I'll simply ask who invented the telephone. So that's the question I'm going to ask, let's see what HAHBD comes up with. Okay, so it looks like for the main part, this is correct. Alexander Graham Bell, he was the main inventor of telephone, where he got the main credit. You can see that in some historical debates, you got Alicia Gray and Antonio. So it's hard to tell right now, but let's pretend that we do know for a fact that Alexander Graham Bell this is correct, but we're not sure about Alicia Gray. Again, for the purposes of this example, let's pretend that this is incorrect because again, it's going to be very hard to try and exactly produce an incorrect response from AI because we have no control over its training dataset, and hence we cannot simulate a specific scenario for the purpose of this SMO. But let's say, for instance, as an example for this scenario, the response introduces misinformation, which is what we see here. Now, what we can do is we can refine the prom for accuracy. So what we can say is use the following prompt and simply say cite reliable sources and confirm who is credited with the invention of telephone. So let's go ahead and run this. And now you can see that it says the person officially credited with the infection of telephone is Alexander Graham Bell. So based on the patent number, so now it's giving you the patent number, the issue of the patent, and the date of the issue, you got the reliable sources. Now you got the patent and trademark, USPTO, you got Labry of Congress and all these valid sources. And again, if you like, you can click on the link here and it will take you to that specific website where you can read all the full details and information. And again, this is talking about some alternative claims, but you can see that initially we got two more people that could have been credited the invention of telephone. But after putting in the second refined prompt or follow up prompt, we can say that for sure, we know it's Alexander Graham Bell. So here, the key takeaway is that always fact check AI generated content, especially for historical or scientific claims to ensure its accuracy. For our next demo, let's take a look at fixing an overly long response. So here, what we're trying to show is that how AI responses can be too detailed and need summarization. And for this walk through, I'm going to ask AI to summarize a complex topic such as quantum computing. So what I'm going to do is I'm going to start with the following prompt that says, explain quantum computing. Okay, great. So now you can see that Chachi PT has given us the results, and broken it down into different categories like classical versus quantum big, superposition, and all these different categories. So this is great, but you can see that the issue or an observation is that the response is too lengthy for a quick understanding. So here what we could do is we can refine the prom to request a brief summary instead. So what I'll be using is a follow up prompt that simply says, summarize quantum computing in one sentence using simple language. Sometimes you see different people using this phrase using simple language. Sometimes you see people telling Chat GPT, explain topic X to me like I'm a 5-year-old or like I'm a 10-year-old. And that's simply saying the same thing. It's saying use very simple language. Don't use technical jargon that I couldn't understand. So in this case, we're asking it for in one sentence, so super brief using simple language. So let's go ahead and run this. And there you go. Right now, you can see that the response is more concise and more clear. So quantum computing is a new kind of computing that uses tiny particles to solve problems much faster by working on many possibilities at once. So again, exactly follow that direction, one sentence, simple terms. Anyone can understand this. So more concise and clear, and you can see that telling AI to summarize or simplify can help make responses more digestible. 23. Improving Consistency: Let's now dive into improving consistency and avoiding conflicting outputs. ChachPT generates responses based on probability, meaning it may give different answers to the same prompt. While this can be useful for creativity, it can also cause inconsistency in structured workflows for professional content creation. AI doesn't remember previous responses, and slight variations in phrasing can shift its focus, leading to different or even conflicting answers. To improve consistency, we must refine our prompting strategies. Consistency improves when we standardize our prompts and add structure. The more precise the instructions, the less variation there will be in AI responses. All right, for the next demo, we're going to be trying and ensuring AI responses follow instructions correctly. And here, what we'll demonstrate is show how different phrasing leads to inconsistent responses, refine the prompt for a more structured than reliable answer and then use constraints on things like such as format, length, tone to show that we can maintain consistency. For the first one, what we are going to do is we're going to test inconsistent responses. And here, we'll show how small prompt changes can lead to significantly different outputs. So here, what we can do is just use the same prompt. So I'm going to be using describe the benefits of meditation. So let's go ahead and run this prompt, see what ChaBT comes up with. Now, we're going to be asking Cha GIPT a general question multiple times. So first, it was able to come up with this. Let's ask the exact same question again. So I'm going to go ahead and rerun the same prompt. Okay, so let's run it one more time. Okay. So if you go back, you can see that the first one, it starts with Meditation offers a wide range of physical, mental and emotional benefits. If you go to the Sekan one, it says something completely different. But it does have somewhat of the same structure. So it starts with the introductory sentence. So it says this one said, meditation offers a wide range of physical, mental and emotional benefits. This one says meditation offers a variety of benefits for your mind, body, and overall well being. It is trying to say something the same thing, but the point is it doesn't remember what it previously said. So it is giving you a variation. So again, going back to probability of expecting what's next in terms of text and output, it is somewhat trying to convey the same message, but the phrasing and wording is completely different. So you can see this one, the first category mental benefits. It is the same for this one. However, you can see that the first one says reduces stress, lowers crystal levels and hormonink to stress. For this one, it says, reduces stress and anxiety by calming the nervous system. So completely different phrasing and responses. And then the same thing for this one, if you're looking at it, you'll see essentially different types of responses. Here, the point of this demo is that to teach you that AI doesn't recall past answers, leading to response variability. So for this next one, we're going to be building on top of what we've already provided, which was described the benefits of meditation from the last lecture. But here we're going to be refining the prom for more reliable responses. So here we're going to show how adding structure reduces variability. So what we'll be doing is modifying the prom to enforce structure. And I'm going to do that by using the following prompt that says, List three benefits of meditation in bullet points with a short explanation for each. So let's go ahead and run this. And now you can see over here, JGBD has produced that, and now you can observe that this response or the output is now structured and more consistent. And here you can see that from observing the results, you can tell that AI provides more consistent responses when the prompt includes structure and formatting instructions. Lastly, we'd like to demonstrate how using constraints using chat GBT constraints to maintain uniformity. So here we'll show how constraints like word limits, formality or response style can improve consistency. So here, you can continue with this chat or open a new chat. It doesn't really matter. But I'm just going to continue with this chat here. But actually, let's open a new chat. Why not? Here, though, you're going to see I'm going to use the following prom to ask AI to follow specific response constraint. And here, I'm going to say, explain the importance of cybersecurity in exactly three sentences in a formal tone. Okay, so you can see now, we're kind of fixing the previous issue we saw from the last lecture so that the response now follows a controlled length and formal style. So here you can see that by setting word limits, such as in exactly three sentences, you can use character limits if you'd like to or word limits, whatever it is, whatever limit you want you can introduce in your prompt. You can see that the word limits, formatting rules and tone requirements, so in this case, we ask for a formal tone, those all can help AI to generate more predictable results. M. 24. Ethical Considerations: So ethical consideration is always a big topic and discussion in the world of AI. So let's take a look at that and consider that in AI generated content. So AI is a powerful tool, but it does not understand ethics or fairness or truthfulness. It simply predicts text based on patterns and its training data and the access it has to the Internet and the information on the Internet. So that's why human oversight is essential to ensure AI generated content is accurate, fair, and ethically responsible. AI generated content can re enforce, biases, spread misinformation or violate copyright laws. So understanding these risks help us use AI responsibly and more ethically. AI content should be verified before being published or shared. By cross referencing reliable sources and eliminating misleading claims, we can ensure ethical AI use in content creation. 25. Creating AI Personas: Alright, now, let's dive into creating AI personas for customized responses. Chat GPT can be customized to behave like an industry expert, customer service representative, or subject matter specialist. By defining its persona, you can enhance AI responses to be more relevant, consistent, and insightful. Defining an AI persona enhances the quality of responses. Whether you need HGPT to act as a financial analyst, doctor or recruiter, shaping its identity makes it more valuable and engaging for your specific use case. To create an AI persona, you define its role, expertise, tone, and response format. This ensures HGPT delivers responses aligned with industry expectations and audience needs. In this next walk through, we're going to be crafting an AI persona for a specific industry. So here we'll demonstrate and start with a basic prompt and observe AI's default behavior. We'll refine the persona with additional instructions, and then we'll test the persona with real world scenarios. So for our first step, let's go ahead and test JAGPT without a persona. So here, I'll show how HAIPT responds more generically without any customization. So here, we're going to start with a very simple request, and I'm going to use the following prom that says, give me advice on investing in index funds. So let's go ahead and run this prompt. And here you can see that HAGPT was able to give us the answer. However, it is somewhat generic. So it does categorize it by understanding what the funds are, start early and be consistent. So this is more of a sort of like an instruction or recommendation. Choose low cost funds, diversify, think long term, and so on. So the response is good. It is still somewhat on the generic side, and here you can see the issue is that the response lacks depth and personalization. So here you can see without a persona, AI provides generic surface level answers. Now, let's build on top of this and define an AI persona for a financial advisor. So here, I'll be showing you how structuring an AI persona enhances its expertise and style. So what we're going to do is modify the prompt to introduce a persona now. So here's what the improved prompt looks like. So what I'm saying here is you are a professional financial advisor with ten plus years of experience. So again, we are defining the persona via the sentence. Your role is to provide expert guidance on investing in index funds. Use real world examples, recommend specific strategies and key responses under 200 words. So we are introducing the word limit here as well. So let's go ahead and run this front. Okay, now we're getting more specific answers here. So the difference here is that the AR response or hATVT response is more refined and is more expert like. So here you can see a well defined persona makes AI responses more specific, informative and actionable. And lastly, let's go ahead and test the AI persona in real world scenario. So here I want to show you how the persona remains consistent across different queries. So for this next example, what I'll do is I'll ask a follow up question to test consistency. So here, I'm going to use the following prompt that simply asks GPT, what's the best strategy for a beginner investing in index funds? So let's go ahead and press Enter, and you can see that it is still maintaining the response is consistent with the persona. So you can see that it's saying for beginners, for a beginner, the best strategy is simplicity, combined with consistency, and then telling you number one, start with a broad market index one, use dollar cost averaging, keep it simple, consider adding bonds for stability, invest through tax advantage accounts, and so on. So here you can see defining AI personas ensures consistent expert like responses across different proms within the same session. 26. Layered Prompting and Nested Queries: Now let's dive into layered prompting and nested queries with JAGBT. Layer prompting improves AI responses by breaking down complex tasks into logical steps. Instead of asking one broad question, we can guide JAGPT through a structured thought process for deeper and more accurate results. Nestet queries allow AI to refine and expand responses naturally, similar to a back and forth conversation. By guiding JAGPT through a logical sequence, we can achieve more detailed and insightful answers. By structuring AI interactions into layers of inquiry, we help Chat GPT generate well organized, thorough and actionable insights. Now, let's go ahead and talk about how we can be using layer prompting for better AI response. And here, we're going to demonstrate by starting with a broad question and observe AI's initial response, we'll use a follow up queries to refine, expand and improve the answer, and then we can compare a single step response versus layered approach. So for the first step, what I'm going to be doing is testing a single broad prompt versus layer prompting. And here you can see how a single vague prompt leads to an overly generic response. So I will start with a very broad request, and I'm going to be using this prompt that says, How do I start a YouTube channel? So let's go ahead and press Enter. So here you can see that it is giving us somewhat of a structured response. So starting with defining your niche, set up your channel, plan your content, get basic equipment, and so on. So you can see that the response is not bad. However, it is still somewhat generic and unstructured in terms of what we need to really be able to create that YouTube channel. So you can see that response lacks details, and this is because with HAGBT, broad questions often lead to shallow and incomplete answers. Okay, so now let's move on to the next step where we will be refining the answer with layer prompting. And here I'll show how breaking down the question improves response steps. So we'll break the task into structured sub questions. So here, what I'm going to do is start with follow up prompt, the first one that simply says, What are the most important steps before launching a YouTube channel? So let's go ahead then presenter. So now you can see it is giving us more detail than structured response. So before launching the YouTube channel, it's essential to lay the solid foundation, and it's telling us what those are. So defining the channel purpose and audience, choosing a niche, research the competition, plan content ahead, create branding elements, set up your equipment, create a trailer or intro video and so on. So this is actually pretty good. You can see that this is actually a lot more detail compared to what we need to get the YouTube channel started. Now let's go ahead and follow up to focus on a key aspect. So for example, the YouTube content. So here, I'm going to put my second follow up prompt. Again, remember we're layering this, right? This is called layer prompting. So now I'm going to follow up with this prompt that says, how do I create engaging YouTube content? So let's go ahead and run this. So it says, creating engaging content is all about capturing attention, delivering value, whether that's education, learning, or whatever, and keep people watching. And this is how you can do it. So hook viewers for the first 15 seconds, focus on one clear message, tell a story, use visuals, encourage viewer interaction, pace matters, watch your analytics, and, you know, see if you can gather some insights from the data that you collect from your viewers. The videos. So here you can see this is more detail and actionable, and this shows that step by step refinement helps JAGPT deliver detail and structure responses. Now, let's move on to the next step where we will be using nested queries for additional depth. So here I'll show how AI can build upon previous responses for richer insights. So one thing we want to do is ask AI to summarize its previous responses. So what I'm going to do is use the following prom that says, summarize the key steps for launching a YouTube channel in one paragraph. So let's go ahead and run this. Okay, great. So Cha GPT was able to do exactly that, summarizing the previous steps into one paragraph. So again, to launch a video successful video channel, start by choosing a clear niche and so on. You can pause the video to read through this. And again, the ha chiPT response is concise and well structured. Now, what I'm going to do is ask HAGPT to suggest an action plan based on the summary. So what I'll do is put in the following prompt here that says, now, create a 30 day action plan for launching a YouTube channel. So this is going to help us going if that's our goal here. So now you can see it's breaking it down week by week and also day by day. So week one, foundation and planning, so it's telling us, again, this is just a recommendation. You can alter this based on your use case or what best works for you and your time schedule or whatever the purpose or use case is. This is just an example where we're trying to start a YouTube channel, but your purposes might be different. So anyways, going back to this, you got week one, so you got day one, Day two, define your niche and write your channel mission and value statement, Day three and four research, day five and six, brainstorm ten video ideas, day seven, choose your channel name, and so on. So this is actually breaking it down really nicely week by week and day by day, providing you an actionable, exactly a 30 day actionable plan where you have a guide, some guidance in terms of what you should be doing throughout the day for each day throughout the week. So you can see here that NSSET queries make AI generated responses more structured, contextual and, of course, actionable. 27. ChatGPT with other AI Tools: Now let's take a look to see how we can make HGPT more powerful by combining it with other AI tools. While HGPT is great for generating text, combining it with other AI tools unlocks even greater potential. Whether you need to automate workflows, analyze data, or create visual content, multi tool integration can significantly enhance productivity. Each AI tool has unique strengths. By combining HGIBT with image generators, spreadsheets and automation platforms, you can create highly efficient workflows tailored to your needs. Integrating HGBT into existing tools streamlines workflows and saves time. Whether through APIs, automation tools like ZapiR or direct AI to AI interactions, HGPT can seamlessly enhance business operations. Right, this next demo is going to be very cool because we are going to be using HAGBT with another AI tool to complete a task. So here we'll start in our demo, we'll start with a task in HAGBT. So in this case, we're going to use data or content creation. Then we'll use another AI tool to enhance the output in this case, image generation, and then you'll see how AI tools work together for a complete solution. So this walk through is going to consist of two steps. In step one, we're going to be generating AI optimized content in HHIBT, which in our case, is going to be a social media post. And then in step two, we're going to be enhancing the content with another AI tool, and for that, we'll be using Canva for image generation, and then we'll bring it all together. Let's start with our first step here, and what I'll show is how CHAIPT generates structured detailed texts for content creation. So for this example, we're going to be starting with a simple request, and we're going to write an Instagram caption or post. So for this, I'm going to use the following prompt that says, write an Instagram caption for a new Fitness Smartwatch launch. So imagine you are wanting to launch a new product and then you want to have a campaign and you want to launch it through Instagram. So let's go ahead and press Enter. And there you go. So HHIPT is giving you the Instagram post. So meet your new work, you'll partner the Next Gen Fit Watch, track every step, every beat and every goal, smarter than ever, and then some hash tags. So if you wanted to, you can go further and enhance this based on whatever tone, audience, and so on. You can also follow the questions or recommendations HHIVT is giving you. So it says one variation for specific audience or tone. So you can define that and then I'll give you a more refined but let's say we're more refined response. But let's say we're fine with what Cha GBT has given us in step one, and now we want to move into step two. Alright, now that we have our content, which we generated using hatchPT which is an AI tool, we want to be enhancing that content with another AI tool. And for this, I'm going to be showing how Canva, which is a design tool with AI capabilities can enhance the HAGPT output and for us to be able to use these tools to devise a complete solution. So here, I'm going to be using Canva AI to create an engaging post. So for this, we're going to be taking the HAGPTGenerate a caption and designing an Instagram post. And there's a couple of ways that I'm going to show you on how to accomplish this. So first of all, before we get started, you just need an account. And again, a Canva account, they have different tiers. They have the paid version, which is the pro and then they have the free for the purposes of this course, and many, many use cases, you're fine using the free tier. So what I recommend is going to the Canva website, which is simply canva.com. All you need is a user name and password. So just use your email and create a password, and they even have login through Google, so you don't necessarily have to create an account. You can just log in with your Google, Apple ID, and things like that. Or you can just simply use an email and create an account. So go ahead and set up your account, and then once you do and you log in, you should be presented with the homepage, which is what it currently looks like. Alright, so Canva has a pretty simple design interface, which is nice. It's very clean. It's easy to use. So on the left hand side, you got your navigation bar. You got your menu. You got your home project templates, brand, Canva, AI and apps and then on the center, you got templates and things like that. So again, it's a design tool for creators and content creators. But you can do other things. Like there's many things you could do. You can create presentations. You can create resumes, cover letters, posts, social posts. There's many thousands of templates available for you to use for free and the paid ones. So great tool overall and highly recommended. But for the purposes of this demo, what we're going to do is we really want to use different AI tools to create an Instagram post for launching our Smartwatch or new Smartwatch product. So let's go ahead and do this one way first. So in order to do this, you can use the AI capability, which is now called CVO AI, and this is something recently Cavo has introduced, which is really. And here you have different options, so you can pick code, write, video, design, image, and so on. And if you click on these things, you can see a sample and even it gives you the prom. So for example, if you wanted to create an image, if you click on this, it will actually start creating that image with this current prompt here. So it says, create an image of a simple skincare bottle with soft botanical shadows in the background. So this is kind of cool. And yeah, this is really nice. And we essentially want to we want to create an image very similar to this, even simpler. But what we're going to do is we need to come up with our own prompt. And again, it really depends on what kind of prompts you come up with. But right now, all you have to do is really we clicked on that image. So let's go ahead and actually backtrack. So if you wanted to do that from scratch yourself without clicking on any of these things, all you have to do is simply make sure that Canva AI is selected here. Again, you got a couple of options. You got your designs. You got templates, and you got Canva AI, so make sure you select that. If it's not selected, and now you can choose the feature from the AI. So you can design for me. You can choose design for me. You can create an image, draft a dog, code, and so on. So for this case, we want to create an image. So go ahead and click that. And one thing is here, you can use different styles. So this is different styles with different filters and so on. So you can choose SMART a cinematic, creative, whatever macro and all these different stock options that's giving you. So for example, we can do something like we can just select none and see what it does. But I encourage you to experiment with all these other ones because they give you some really interesting and realistic results. So for now, we'll leave the star as is, but Instagram posts are typically not with a 169 Asterk ratio. They're the opposite, so they're actually 916. So we need to change that to make sure our image fits the dimensions correctly on the devices that people are using to go on social media platforms like Facebook and Instagram on their phone. So now aspect ratio is changed to 916, and now all we have to do is put in our prom. So in our case, we used we were talking about a smart voice, so we simply want to create an image for a Smart Voice. So all I'm going to do is say a simple brand new smartwatch in the background. So let's go ahead and press Center and see what Canva comes up with. Okay, awesome. So Canva has finished creating four different variations. And again, this is by default. It has the aspect ratio correctly, 916, as you can see, it's giving us four different variations of a Smartwatch. So this is actually really nice and you can start you can start your design with this as your baseline, and then you can layer up onto it. Now, all of these look great. This one, the text is a little bit messed up, so we're not going to be using that. This second one and the fourth one look pretty good, in my opinion. So we can use either one, doesn't matter. But if you go here, there's a couple of options because now this is your baseline. What you could do is could do many different things. You can just download this and again, take it back to Canva and start designing and layering things up with text, logos and stuff like that. You could take it to another you can download the image and upload it to another software like Photoshop and edit it there. Whatever you desire, you can do that. So on the bottom left hand side, you got this arrow button, which is the download. Over here, you got the dot dot dot, which is a setting so you can copy and delete the image. And of course, you can click Edit, and this is going to just use Canvas editor. That's built in that you can leverage, which I highly recommend. So either the fourth one or this one, let's just go with the second one here. This one looks very simple, elegant and nice. I also like this one because it's simple. It's got the shadow and everything very soothing colors in the background, same with this one. So now we have our image, and we're satisfied with this. If you're not satisfied, you can always go back. And over here, you can put in a different prompt and just experiment. And note that on the right hand side here, this is the available token that you have with the free accounts. And if you wanted more, you could simply upgrade, and Canva even gives you a pro trial for 30 days, so you don't have to commit to anything. And if you like it, then you can continue the subscription, which is paid. But for now, the free tier is good enough. So far, we have enough tokens, but we're done. We are happy with this photo, so let's go ahead and click it. And this should open the Canva editor, and this is where you can go ahead and experiment and start layering up and building on top of the foundation of your base image, which is the smartwatch. Now that we have new editor, the image loaded in our Canva editor, we can start editing the image and just build on top of it. So you can see by default, it opens the image tab, and there's many, many things you can do here. You got magic Studio. Most of these, unfortunately, are Peto, so you can see this icon with the crown here, the yellow crown, that means paid. But there's many different things you could do. So you can apply filters here. So, for example, if I change Fresco, you can see that will change Belvedre and so on. So this is All nice. Let's just go back to none for now, and then you can apply effects like shadows. You can apply you can use different apps on here. And yeah, there's so many different things you could do. Now, for the purposes of this demo, the goal is not to learn Canva here. They're editing and image creation. This is just how we can use AI tools to bring things together. So all I'm really going to be doing is just adding text now. So I'm happy with this image. Let's say I've done all my editing. I'm happy with this. So all we need to really go into the text one, here, there's many different text things you can use here. Again, different different styles, different fonts, different colors, so many variations you can do. If you want to play one, just click on AD a Textbox, and then you can start editing it from there. You can choose one of these ones. So for example, I'm going to choose this one, empower your team, and then we could do something else. Let's just do a regular one. Let's just go add the textbox here. And then when you add a textbox, let's go ahead and push this down. Okay, so we got a couple of texts. These are placeholder text, obviously. And then, you know, when you have this, you can change the font and the size and the style and effects and things like that. So we're happy with what we have right now. So all we need to do now I want to complete my Instagram post. So I got my image. I just have to put the text that we got from ChaiPT and then we have a complete infographic and ready to post on Instagram. So here we could do this. We could just go do something like let's just copy paste this whole thing there. Let's go back and paste that in there. Okay, and the colors you can play around with this. This is not that great. Let's if we go to colors here, you can change it to whatever you want. Change it into a little bit of a darker purple here. You can experiment with these colors. The blue one is not bad either. But again, it really depends on your design and preference. So let's say we're happy with this, we can read what it says, meet your new workout partner, the next fit Wash. So let's say we're happy with this again, not that great. You can experiment with this to find what works best for you. But yeah. And then what we want to do is go back and then grab the rest of the text. So, let's say, for example, this one, and then let's go ahead. Let's paste this in. The text for this is too big, so we're going to be reducing the size. Let's go to 21 bit too small. Let's go 32. 36. Okay, this is not bad. So we'll bring it down, so it's readable. And again, if you wanted to, you could change the color by just clicking on that. And then there's all the things that you could do here and change the size, change the font, change the color. And yeah, there's many things you could do with this. So now you are done with this. The image is ready. You got your smartwatch, you got your text and caption. Now, all you really have to do is download this and then go to Instagram post it and post your hashtags and hopefully that'll kind of go viral. So this should help not promising anything, so it may or may not go viral. But again, the point of this is to teach you how to use different AI tools to create a solution and to add. And now that this is already, what you can do is you can just simply click Share, and then you can click on Download, and then this will download the image. There's several different settings you can choos. So by default, it's the PNG, which is the image format. You could do PDF NP four, GIF, PT, PowerPoint, and JPCFimages PNG and JPEC are good. So you can go ahead and select that and then click on load, and that will take a few seconds to process that and then it will download it to your device. So you can see this is how you can bring different AI tools to work together to complete solution end to end. And Canva is a great design tool in conjunction with HAGBT. And here you can see throughout this demo that pairing HAGBT with design tools such as Canva makes AI generated content visually very appealing. Now, what I would like to do is show you a couple of different ways using Canva to accomplish the same thing just so that you get a sense of how powerful these AI tools are. Now, what we can do is go to Canva and then you can select Canva AI here. You can select design for me or create an image, or you can just leave things as is. And you can see the placeholder text here, similar to Chachi BT or other AI tools, you can actually chat or conversate with the CanvaI through prompting. So here it says, describe your idea, and I'll bring it to life. So a very similar feature set. So you can speak to it. If you use the voice or the microphone icon, you can add media such as files and folders and things like that. But here, we're just going to do a very basic prompt. But we're still keeping in mind the example that we're trying to accomplish, which is the new Smartwatch launch. So here, instead, what I'm going to do I'm going to try to get it to do majority of the work for me, so I'm just going to say, create an Instagram post that promotes the launch of a new smartwatch. And then you can go one step further and you can actually say the caption for this post should include, and then you can copy in the one that we got from HAGPT. So if you go here, for example, you can really pick any text or create your own text if you like, I'm just going to copy the one we got from CHAPT. Go here, and then I'm going to put that in codes and paste that in. So create an Instagram post that promotes the launch of a new smartwatch. The caption for this post should include and then this text. So let's go ahead and run this and see what Cava comes up with. And it says, I just needs a few minutes to put this together for us. You can see it's creating various options. And there you have it. So Canva finished putting together some variations here of this prom. So now you can use any of these that you like for your Instagram post, and you got some reels here, which is quite nice. It can add its own text and infographics, which again, is quite nice and useful and handy. And you got some variations here so you can use whichever you like. So this one is kind of cool here with nice colors and theme and everything. So choose either, you can click on the edit, take this to the Canva Editor and further edit this to add more objects or change colors and things like that. You can continue the conversation in prompt and change somebody's. You can even click on more design. You could change your prompt with pictures of athletic women running smartwatch. So you could do a lot of things with this, and this is a good starting baseline for your design and the promotion of this new Smartwatch launch. Let's take a look at one more way that you can create this engaging Instagram post. So now we can go out of the Canva AI, so we can exit out of this and actually create templates because Canva has thousands of templates that you can use, and this is pretty up to date with today's trends. So you can see they have Instagram posts already here. If not, you can just type in. They actually have millions of templates. So you can type in at Instagram post, for instance. Let's go ahead and click on templates on the top. Then let's go ahead and click Instagram post. And then this is where you start getting sort of all these templates that you could potentially just select from. Or you can just type in what you want. And you can also determine your style here. So modern minimalist, simple, elegant. So I'm going to do elegant. And then these are some of the ones that you can choose from, right? So let's go ahead and type in Smartwatch New smartwatch. And then here you can search for Smartwatch. And then here you get all the templates with the Smartwatch that we could potentially use as our baseline. Now, please note that a lot of these are sort of come with the paid subscription, so you can see the crown yellow crown icon means paid. So you'll just have to scroll and see if you can find one that it's free that doesn't have the crown. So we'll change this style to all style, see what we come up with and see if there are some free options here. Okay, so you got a few free options here. Again, it really depends on what you need. You can also even start with a blank one, but let's go ahead. I just wanted to show you this. So let's just use a free one here. So, for example, this one, let's go ahead and click this. You can say customize this template, and then you can click on this, and then this will take to the editor and use this as your template. So now, what you could do is again go back to CHA GBT, copy paste this text, and then paste it here and this, and now you got another Instagram post. Or potentially you could again, put any text of your choice, and you could do change the colors, font size, bring in objects, and so on. One last thing here I wanted to show you is that you could generate your images through a different means or entry point and not necessarily CanvI. So for instance, let's say in this Instagram post or this image, we don't really like these smartwatches. So we want to generate our own. So let's go ahead and select these and then remove them. And what you can do is come up over here to elements. And then if you see there's a lot of things there's shapes, there's graphics. And what I would like you to do is take a look at the image generator. So there's already stock photos and things like that, but let's go ahead and the image generator and it says, generate your own. So let's go ahead and click this. And then on their images, you got images, graphic and video, we're just going to stick with images, and then I'm just going to put in new smartwatch, three D smart watch. So let's go ahead and run this. And now Cama is going to through magic media feature, it's going to create this option for us. Okay, so there you go. It generated four different variations. They all look pretty good, so you could choose whichever you like. Again, you can play around with the prompt, put something else. You can play around with styles, the aspect ratio, and what you like. But let's say, for example, we like this one. So if you click this, it's going to bring this in, and now you can sort of go ahead and play around with this, and this will be your new Smartwatch. Again, there are some features. For example, the background remover is a cool feature, but it is paid. There are other free tools you can use to remove the background, but let's say for the purposes of this example, we're happy with this, and this becomes our final Instagram post. But again, there's many different feature sets in Canva that can help you generate really, really cool infographics for social media posts. 28. Automating Workflows: Although we won't be going through a lot of depth and detail in terms of connecting and integrating with HGT API and other applications, I just wanted to briefly touch on this so that you are well aware that it is possible to automate workflows with HGBT and APIs with third party applications. Automating workflows with HGBT saves time, reduces errors, and increases efficiency. By integrating AI with APIs, businesses can streamline operations, improve customer engagement, and reduce repetitive tasks. HatGPTs API allows developers to integrate AI into business applications, websites, and workflows, whether through code based implementation or no code tools, AI driven automation enhances operational efficiency. AI automation can be applied across multiple industries from marketing and sales to customer support and analytics. When combined with APIs, AI acts as an intelligent assistant that simplifies complex processes. 29. Practical Exercise: It's now time for a practical exercise where we create a personalized AI persona for a use case. In this exercise, you'll create a custom AI persona designed for a specific use case. Whether it's for customer support, coaching or technical consulting, shaping HACEPTs responses will make AI more relevant and valuable in your everyday workflow. Think about which industry or function you want your AI persona to specialize in. Clearly, defining its role and expertise ensures more precise and useful AI generated responses. The tone, response structure, and detail level of your AI persona should match its function. For example, a legal assistant should use formal language while a social media strategist might adopt a casual, engaging tone. Here you want to see how the AI persona remains consistent across various interactions. You should observe after this exercise that defining AI personas ensures consistent expert like responses across different queries. AI personas can be applied to content marketing, education, tech support, and customer service. Customizing AI responses improves accuracy, engagement, and workflow efficiency. 30. Future of AI and Prompt Engineering: Now let's spend a little bit of time talking about the future of AI and prompt engineering. As AI continues to advance, prompt engineering will become more sophisticated. AI SEP Stems will better understand contexts, refine outputs based on real time learning, and integrate more deeply into business and personal workflows. AI is moving towards better memory retention, multi modal capabilities, and creative assistance. As models improve, AI will become more human like in its interactions, improving efficiency across multiple industries. In the coming years, AI will transform industries by optimizing workflows, automating repetitive tasks, and enabling smarter decision making. Businesses that integrate AI early will stay ahead in innovation and efficiency. 31. Bonus Tips and Resources: Let me share some resources to help you for your continuous learning and AI journey. AI technology is advancing faster than ever now. Staying informed about new models, prompt techniques, and AI applications ensures you remain competitive and get the best results from AI Power tools. The best way to keep up with AI advancements is by following trusted AI research blogs, trend reports, and hands on courses. Many of these resources are free and regularly updated. AI skills improve with regular experimentation and real world application. Practicing with PMs following industry trends and working on AI projects sharpens your expertise over time. 32. Recap and Next Steps: Let's now go through the next steps together. But first, I'd like to take a moment and say, congratulations on completing the course ha GBT prompt Engineering. You've gained deep insights into AI prompting, workflow automation, and AI powered decision making skills that are in high demand across industries. You've transformed from a beginner to an AI power user. By mastering structured prompting, AI automation and workflow integration, you can now leverage AI for productivity, creativity, and efficiency. Whether you're in marketing, tech, finance, or business, AI can enhance productivity, automate tasks, and streamline decision making. The next step is to apply these skills that you learned in the course in real world scenarios.