Mastering Generative AI: From Beginners to Advance with Prompt Engineering

1. Introduction ChatGPT Beginners course: Hello friends, welcome to the first new lesson. I have been teaching you about Lean Six Sigma. And I have been teaching you about storytelling with data. When I was talking to one of my friends, I understood that it's important for us to know what is chat activity because when I was talking to my friend and told him that do you know about Changi PT and he was blank. And that's where I felt that all my students should also be knowing about chat to me. This course is for beginners. This course will help you understand what is the activity, what is the role of AI? And you will see the art of possibility that when a man meets with an intelligent machine, what all can be created. Do not worry. Do not have the fear that you're going to lose your job. In fact, people with less capabilities will also now become employable because of the presence of charging. So it is a new possibility for all of us. And I felt that it's very important for us to understand and use these to the fullest potential. It is like a small assistant that you will have who is invisible but always at your service. I think I should call him like a genie. You make a wish and it will fulfill your command. If as long as it is in text as of March 2023, I would not be surprised that if Chad zippity will get more smarter to respond to your audio messages and to your video conversation. But whatever I have used in Chad DBT, by typing, I have got the most appropriate response. Most of the time. It isn't a continuous evolution state. So there is a lot of changes which will happen when you might be seeing this course. Because on a daily basis, Open AI is doing a lot of research to making it more flexible, more adaptable, and more human-like. When it responds to your questions. This is an amazing evolution that we have made as humans because at the end, jagged GPT is also created by humans. But we are using the possibility of machine, which is learning from its mistakes. Let's understand more by diving into this problem. See you next. 2. What is AI : Hi, everyone. Today we are going to tackle an important topic. What is AI? The word artificial intelligence. You have probably heard about it. AI is in conversation everywhere and it is seen in action with tools like Exa, Siri, and even our Netflix recommendation. But what does AI really mean? Don't worry, by the end of this class, you will have a clear understanding of AI. Explain in a way that is super easy to grasp. Let's dime in. All right. Let's start with the basic. AI stands for artificial intelligence. In simple terms, it's when the computer is programmed to think and learn like humans. Think about all the things that you use your brain for. Example, you are solving math problems. You are recognizing the faces of your friends and families and relatives, understanding the language, and even deciding what to eat. AI tries to replicate these abilities using the power of computers. Let me give you an example. Imagine you are using Google Maps to find the fastest route home. Google Map doesn't think like a human, but it uses AI to process traffic data, calculates travel time and give you the best route all in seconds. That is AI in action. Now, let's break it up further. Computers, even the smartest one, only understand one language. That is zeros and ones. What we call as binary code, everything from videos to pictures to text has to be converted into zeros and ones so that the computer can process it. For example, when you type the word Pou Badji recipe in a search engine, the computer doesn't actually understand the word. It converts inputs into zeros and ones. Searches for similar pattern in its database and shows you the results. Sound school, right? But there's a catch. Early AI systems were limited because they couldn't understand the deeper meaning behind the words. Let's take one more example. Let me explain. Say you search for the word cat. A simple computer might think cat is similar to bat because both words are short and have similar letters, but it won't understand the cat and the kitten. They are related because they represent the same animal at different life stages. This is where modern AI steps in. It uses something called as vectors to understand words in a smarter way. Think of vectors like a report card, just like you get scores in subjects like mathematics, science, and English, AI understand word scores across hundreds of categories. For example, the word cat might score high in categories like animal, pet, and furry. The word kitten might have a similar score which helps AI to understand that these words are connected. This scoring system allows AI to figure out meaning and relationship between words, making it much smarter than the simple matching based on letters. Let's make this even simpler. Imagine you're shopping for a pair of snickers. You don't just care one feature. You want shoes to be stylish, durable, and comfortable. Similarly, AI assigns hundreds and even thousands of scores to the words, phrases, and the concepts to capture their true meaning. For example, tools like chat GPT use over 1,500 dimensions to understand the language. That's why it can answer your questions so efficiently. It understands not just the words, but also the context behind them. Now let's talk about another powerful technology behind AI that is neural network, artificial neural network. These are inspired by the way our human brain works. Our brains have billions of neurons that help us think, remember, and solve the problems. A neural network is like a digital brain. Instead of neuron, it has tiny memory units, and these are callless parameters. These parameters are connected in a huge web like format, just like the connection in our brain. When you feed information into the artificial neural network, it processes it step by step, learning patterns and making decisions along the way. Here's a fun example. Imagine a neural network is trying to recognize Wada Pam. It starts by looking at basic shapes like the circle, the bun, and the texture, the crispy vada. Then it combines the feature to figure out that, Hey, this is a Wada paw. So what happens when you combine these technologies, vectors and neural network? You get advanced AI systems. These are called large language models or LLMs. These are the brain behind tools like Siren Alexa, and ChaGPD. Here's how it works. Let's say you type, what is the capital of Mastra into an AI system. It breaks your question into vectors, processes it through neural network, and predicts most likely answers based on anything it has learned. In this case, it will tell you the answer as mumbi. Let's bring it back to the everyday life. AI is everywhere. Here are a few examples. Voice assistant, when you ask Alexa, what's the weather today? It uses AI to process your voices and understand the question and fetch the answer. Streaming services like Netflix uses AI to recommend shows. If you've been watch a thriller, Netflix learns your preferences and suggests similar movies. Online shopping. Platforms like Amazon uses AI to recommend products based on what you have browsed or purchased earlier. These are just a few ways. AI is quietly making our life much easier every day. But let's not forget AI is a tool, not magic. It works because human design train and guide it. For example, when creating a voice assistant, developers teach it how to recognize different accents and languages. And while AI can seem to be very super smart, it still has some limitations. For instance, it can't think creatively or feel the emotions like we do. The gap between the human intelligence and the artificial intelligence is still big. But thanks to the technologies like vector and artificial neural network, AI is getting better at solving problems, learning on its own, and understanding the context and making the prediction. AI has moved to the next level of self learning. To sum it up all, AI stands for artificial intelligence. It's when computers are programmed to think and learn like humans. AI uses technologies like vectors to understand, meaning neural networks to process complex patterns. We see AI in action in every day from Google Maps to Netflix to voice assistance and online shopping. I hope this has helped you in understanding what exactly is AI and how we can enjoy using it and not get boggled that Oh, it's a technical concept. So we will continue in the next lesson. Thank you. 3. Use-case: Let's dive into the program of Chad ZPD. This is a complete course for beginner. The agenda that I have for you is that I want to introduce you to chat GPT. If you have not heard about it, or you just have a vague idea about somebody talking or with chat activity over a cup of coffee or in a beer bar. What is Open AI and what is its role in getting charged UP into our life? We will be understanding many examples and I'll be practically taking you through lots of learning object is to understand the fundamentals of Chad GPT, learn about OpenAI, how natural language processing, it's not neuro linguistic programming, but how natural language processing, artificial intelligence will benefit with the presence of jagged DPD or with the introduction of captivity into our life. How can I set up chat GPD and Open AI account? I will cover that too. Let's take a scenario. How can I be leveraged? So we understood that chat GPT is an artificial intelligence board which is generating responses to your questions. How can I leverage it to my benefit? Let's try to see. Let's take this example where I have Mr. Ram, who has just moved recently to Germany to pursue his master's. He is having difficulty in conversing and gentlemen. Chris is his roommate and in Germany. So as soon as he sees Ram, He says, Hello Ram and he speaks in German. Brown does not understand what did he asked? Then Ramsay's yes, I'm new to Germany. Is there any easy way I can understand joke. Let me guest. You don't know German? Yes. I don't know German. So let's take the help of LGBT. It will make your life easier. Chad GBT, What's it? It will help you that you type few commonly used German sentence and then English translation. Then it is giving you examples on the screen, which tells you that, okay, How does I learned German? So I do not have to go and buy any books. I do not have to go and do any Google. Can you tell me more about this? I will explain everything that chatbots to chancy beauty. Let's get started. Shall we? I will continue this in the next chapter. 4. A greater level of understanding. awareness. insight.: Well, how did chat typically come into existence? As you know, that there are a lot of global sensation which you will find in mult. Multiple newspaper articles which show that how Chad DBT has come into existence. It gives rise. It's an AI tool which helps X-Man to a next level. I know some of the news articles tell us that it is going to create more of a layoff. But it's not like that. It's like an intelligent machine which will help you work. Let's understand this in more detail. Earlier we had, had bought, I bought is a computer program that is designed to simulate conversations with human users, especially over the Internet. The goal of a chat bot is to provide a quick and automated customer service to help customers find the information they need. Chatbots can be integrated with websites, messaging platforms, mobile apps, and even standalone machines. There are different types of chatbots which are available in the market. Some chatbots are rule-based chatbots. They follow a set of rules and decision trees and determine the response that the user has to. If you go to any of the Indian bank website, you ask a question and it responds based on our own. There are some chatbots which has self-learning chatbots. These are AI and ML algorithms to understand the user's intent and provide appropriate response. You would have seen some boards which respond on Facebook, on WhatsApp. These are continuously learning and evolving. Some chatbots are task-oriented chatbots. They are designed to performing specific tasks, such as booking a flight or ordering food. You also have some conversational chatbots which simulate a human-like conversation. Siri, Alexa, that you say, Hey Siri, what's the weather outside and Syria response back. It's a type of chat bot which has a conversation with you. They will also have a hybrid chatbot, which is a combination of a rule base plus a self-learning technique to provide a more advanced and a natural conversation expert experience. People who have used city earlier versus in the latest phone, they find that Siri is constantly evolving. This is because it's evolving in terms of the content and it's learning and making it more natural. You feel that you are talking to a girl on the other end, instead of talking to. That's how the chatbots have inward. So how can generative AI? Generative AI is a type of an artificial intelligence machine that creates new and original contents such as text, images, music based on a given set of inputs or rules by the human. Ear is trained on a largest dataset and can generate new contents similar to the data that it is trained on, but not an exact copy. So it is generating based on it. What is the history of GPD model? Let's try to understand that in the next chapter. 5. AI Strategic Solution or Problem: Welcome. Today, we're going to talk about AI, is it a strategic solution or a problem that we are facing. Artificial intelligence, as many of you are excited about, is the buzzword that's going around. Gen AI is one more term that we are constantly hearing. I'm sure many of you are already excited about it just like the way I am. However, in my experience both in my environment and during my coaching session with companies large and small, I have noticed that people either love AI or they struggle with it. This polarization can make adoption challenging. As you know, how difficult it is to adapt to the Internet and mobile phones. I still remember the skepticism around mobile phones in 1999. People would say, why should I go for a mobile phone when I have an answering machine at home, or if someone wants to reach me, they can write me a letter. There was a time that we also had pages, which would buzz with the SMS, and people were quite comfortable with it. It is just 25 years ago and see how far we have come. AI has rapid growth reminds me that those earlier days, the speed at which AI is advancing is both thrilling and daunting. Some embrace it while others resist it, simply because as we are humans aren't always ready for change so quickly. But the impact of the AI is inevitable. And you cannot deny how it is going to change our lives for a positive change. Before we dive into the session, I would like to share three key points with you. First, I will reflect on the proximity of AI in our daily lives. Drawing on examples mentioned earlier. Well, we'll explore its implication on business. I will provide some practical steps for implementing AI in your own work. I won't focus on specific tools, but rather on how we can start integrating artificial intelligence into your strategies. I'll also conclude with two quotes that I believe will inspire you as it has done to me. L et's get started. Change or implement the AI as a technology? AI is advancing rapidly, especially with tools like GPT. It's crucial to adopt quickly, However, with such a fast change, there is an important ethical consideration we need to address. One that is being discussed not only in Netherlands, but globally. One major concern is the potential of AI to be misused. Imagine someone creating a video where it looks like a public figure is seeing something. They would never say. For example, what if someone made a video like this? We're entering an era in which our enemies can make it look like anyone is saying anything at any point in time, even if they would never say those things. So, for instance, they could have me say things like, I don't know. Kill Monger was right, or Ben Carson is in the sunken place. Or how about this? Simply, President Trump is a total and complete dip ****. Now, you see, I would never say these things, at least not in a public address, but someone else would. Someone like Jordan Peel. This is a dangerous time. Moving forward, we need to be more vigilant with what we trust from the Internet. That's a time when we need to rely on trusted new sources. It may sound basic, but how we move forward and the age of information is going to be the difference between whether we survive or whether we become some kind of ****** up dystopia. Thank you. Stay woke pitches. One major concern is the potential of AI to be misused. Imagine someone creating a video where it looks like a public figure is saying something. They would never say. For example, what if someone made a video like this of a well known actor or a politician? It said something that was completely out of character like insulting another person or endorsing on controversial opinion. This technology exists today and is called a D fake. It's become harder to tell what is real and what is not. These kind of scenarios show the power and the danger of AI. As we move forward, it's more important than ever to be careful about what we believe online. We need to rely on trusted source and be critical on what we see and here. In this new era of information, how we manage and implement AI will make the difference between the bright future and one where misinformation leads to astray. This is why it's so important to understand not just the benefits of artificial intelligence, but also the risk associated with it. Today, we'll explore some steps that can help you implement AI in a responsible way in your business and navigate it challenging landscape. Let's talk about how AI is affecting our perception of reality. For example, there is a recent buzz about a beautiful holiday destination that many people got excited about. Even tried to book trips too. That's the only problem. The site did not exist in physical. It was a stunning image created by AI, completed with waterfalls and scenic views, but it was completely fake. This shows how easily AI can create things that seem so real and natural, but they aren't existing in reality, which can lead to confusion and even disappointment. We need to be aware of these kinds of situations and understand the difference between reality and AI generated content. On the other hand, AI is also pushing boundaries in creativity. Take the example of an artist who developed a prop for three AI powered artist. He gave the same picture and asked each of it to create a unique painting. Each AI artist came up with a different artwork. Then the artist took it into a step further by asking the AI artist to evaluate the work of their own creation. This example highlights some incredible work. A human artist interacting with an AI artist in a creative process. It's amazing but also a bit mind boggling. To think about AI being this close to us in the realm of art and creativity. We all believe that AI can do manual and rule based task. But with the invent of gen AI, it is even able to think creatively just like the human did. This is just one way AI is getting closer to us and blending into the area, we used to think were purely human. AI in everyday life, some simple examples. I want to share some everyday example of how AI is already a part of our lives, even in simple things like brushing your teeth. For example, I use a toothbrush with AI built into it. Every day I brush my teeth in for 2 minutes, and the tooth brush connects to an app in my iPhone. The app shows me how well I'm brushing and even gives me tip on how I can improve my brushing techniques. This is almost like having a personal coach on my dental care, right in my own home. This is a clear example of how AI is becoming everyday tool and not something far off or futuristic. People often say AI is new, but that's not entirely true. Another example is a tool I use on Linden, which is called a crystal. This AI tool analyzes behavior and personality traits. For instance, it tells me that one of my connection is Len, is a steady and a calm person. It also analyzes my profile and suggests that I'm a bit more dominant and influential. This current of AI helps me understand people better and tailor my communication with them. These examples show how AI is quietly integrated into our daily routine, helping us in a small but significant way. It's not just about complex technology, it's about making our lives easier and more efficient. When we talk about artificial intelligence or AI, it's important to remember that at its core AI is a statistical tool. This means that AI relies on a large amount of data to make prediction about what might happen in future. For example, about ten years ago, there was a headline in Google that could predict the time of your death. This sounds shocking, but this prediction was based on AI analyzing large amount of data. AI noticed patterns, like if people are searched on a specific health related terms. It sounds shocking, but this prediction was based on AI analyzing large amount of data. AI notice patterns, like if many people searched for specific health related terms, and after some time, those search stop. Based on this data, AI could predict outcomes, their potential health issues. This shows that AI isn't new. It's been around for a while. It's deeply rooted in statistics. EI works on analyzing vast amount of data and identifying patterns, but it's most important to remember that these are just predictions based on existing data, not absolute truth. While AI has been incredibly existing and offers powerful insights, we should always be aware that there is fundamental about data and statistics. This understanding helps us to see both the potential and the limit Understanding AI's impact on our lives. Artificial intelligence is becoming more advanced every day. For example, some robots can now recognize human emotions and even detect if someone might be feeling suicidal. However, due to ethical and privacy concern, these robots aren't allowed to share this information directly with a doctor. This shows how complex and fast AI is developing, and it makes us think how we can use technology in our environment. Consider platforms like Spotify and Netflix. They use AI to suggest music and movies to us based on what we have watched or listened to before. This is convenient because it feels personalized. But it also means that we don't all see the same recommendation. AI learns from our habits, which is both fascinating and sometimes we need to be aware of. Another example of a smart washing machine like Siemens. This machine uses AI to suggest the best way to wash your clothes. Optimize the process based on intelligent sensing. AI is all around us. In ways we might not even notice. While AI can be very exciting and useful. Like in case of Smart to Prush, personalized music playlist. We must also be cautious about it. AI has many benefits, but we should always be mindful of what it's doing and how it will affect our lives. I also want to touch on how AI can be integrated into business in the next slide. I will present six key success factors for using AI in business. One is helpful tool in business model Canvas by Alexander Oster. Many of you might be familiar with it. This model can be used to analyze how AI can fit into different aspects of business, such as customer segments, value proposition, its useful way of thinking about reinventing business strategies with AI. In summary, when AI can greatly enhance our daily life business operations, it's crucial to remain aware of its potential impact and challenges. The six key issues to consider when integrating AI into your business. I want to share six important points with you today as we talk about generating ideas in the context of AI. Da generation. This is a creative process where you need to think outside the box. When thinking about how to use AI in your organization, it's important not to get stuck in old ways of doing things. AI is here to stay and we need to adopt it, just like we had to adopt when Internet came in our lives 30 years ago. Prioritization. You can't do everything at once. It's crucial to make choices and prioritize. Focus on the most important idea and develop them into a solid business case. Developing a business case. Many people forget to fully develop their business case before starting a project. This step is vital. You need to understand the value of AI can bring to your business and plan accordingly. Building a test team. In every organization, you will find both positive and negative viewpoints. When forming a team, it includes people from both the sides. This will give you a balanced perspective and help you in making well rounded decision. Fifth point. Small steps and experimentation. Start, experiment with your design, business idea, and AI implementation. Take small steps and learn from them and continuously improve. Sixth. It's an iterative process. The process of generating idea, testing, and learning is iterative. It keeps going after each step. Go back and refine your ideas. This cycle of improvement is essential for successful AI integration. Remember, AI will only continue to grow. Instead of resisting it, think about how you can leverage it to benefit your organization. Let us now talk about integrating AI into your business and career, a practical example. Let's imagine a scenario where you are using AI in a platform called A or ABC. This platform is designed to match job seekers with employers. The AI takes information from both the parties and tries to make the best possible match. The idea here is that the AI can speed up the process, making it faster and more efficient than doing it manually. However, we need to remember that even though AI can help, the human touch is still crucial. Sometimes the AI might not get in write, that's okay. When it doesn't work, we will learn from it, adjust our approach and try again. This is why the process of continuous leaf working or iteration is important. You keep generating ideas, testing them, and refining them. For example, there's a website called there is an AI for that.com, where you can find thousands of AI applications for different tasks like content creation to study aids. But if you try to use new AI application every day, it would be overwhelming. Instead, you should choose one, test it thoroughly and see how it fits into your business. After testing, you can decide whether to continue using it or try something else. Instead, you can choose one, test it thoroughly and see how it fits into your work. After testing, you can decide whether to continue using it or try something else. Let me share a personal experience. I was exploring AI applications for creating courses. I found tools that help me design a marketing campaign. And a marketing course. The results was surprisingly good. Maybe even better that I could have done on my own. However, while the AI was great generating ideas, I still needed my inputs to tailor the course for specific needs. This shows that while AI can enhance my work, the human judgment is still necessary to refine and perfect the results. Next, when you develop your business case, think about the value the AI application will bring. Ask yourself, what problems will AI solve? How will it improve my work? How will it improve my business? Consider both the cost and the potential benefit. Remember, just like the way Internet didn't replace people, but created new opportunities, AI will also do the same. It might change the way we work, but it won't make us obsolete. For instance, by using AI tool, I mentioned I save time on creating creative thinking. And planning for my course. This allowed me to focus on refining the content, ensuring it is met our specific goals. It is also about finding the right balance between leveraging AI capabilities and applying human insights to achieve the best results. 6. Openai Account setup: Igbt, as we're going to go and join this program on chancy PT Bootcamp. It's important for you to create your login ID and password. So I will take you through the process in a step-by-step instruction. So let's get started. As you have heard a lot about tragic beauty, you might, some of you might have already created an account which Ltd. For those who did not create it, your other instructions. At a high level, these are the steps that we follow to create our account and manage our chart. So you will go to Google and type opening a chat DBT. You should ensure that the website that you're not going in has the URL open, the i.com chat, typically, it will be a green color screen where you are going to type chat GPT. I click on the link which takes you there. Once you do that, you will get a page which looks like this, which will say that do you want to login or sign up? I believe if you've already created your login ID, then you don't need to follow the steps. But if it's a new account and click on, Sign up. A small dip over here, is that it will always give you an option to link your Google or Microsoft account with it, which is perfectly safe. But personally it's not my choice. So what I would do is I will provide an e-mail. I do a personally main idea, either a Gmail or Yahoo me already. If mail or whatever email you follow, You can give you an e-mail id and click on Continue. It is perfectly safe to continue with Google or Microsoft account. But personally, I do not prefer because I have creating an account through any molality, I need to enter my password. So I will create a password which is easy for me to remember. It should be a combination of minimum eight words, characters. It will be a combination of alphabets, numbers, capitals, small letters, and so on. So once I have created my account, it will ask me to go ahead and verify my e-mail address. You would receive an e-mail from OpenAI. You will have to click on this button, authenticates that this email address belongs to you and nobody else's opening an account because they have your email ID. And once you verify, it would say that the link has helped you connect with chat. Typically. It's not all one too. I've done that. It will ask you to connect with, but verify your human verification plus a mobile number because chat GBT doesn't want multiple accounts to be created for the same num percent. Once you click on this, you will get an Apache screen telling them identify the traffic light, identify the cross roads, etc. Once you finish that, it will ask you to give your phone number. Only one account can be created with one phone number. Be careful of what phone number you are using. It's only a onetime authentication. It will take once you give it. And if you say Do you want it as a WhatsApp code and you click on X, you will receive it as a WhatsApp message, or you will receive it as an SMS code. Whatever code you have received, it would be a six digit letter, which you will be entering over here and verifying your account. So two things OpenAI does is that it verifies your e-mail ID. Once the e-mail verification is done, it verifies your phone number. Do not worry, they're not going to span your phone number with any of the messages. It just to ensure that one person has only one. An important note, never, ever create your login ID with your office e-mail ID. Like don't create it with other data models.com or the red cognizant.com or so on. Please ensure that you are using a personal meeting ID only for creating this account from a long term point of view. It will ask you that, okay, we are going to give you this. Please read all the instructions that come on the screen. And you are, why are you using? It's completely optional. If you want to click that I'm using it for personal use, or I'm currently created or I'm a researcher and so on. Your choice of selection does not influence the WeChat, GPT works with you. Let's go ahead and select whatever you like, or you can skip this question. And you can continue with Europe known. As soon as you open your account, your account will look something like this. They will give you three examples. What are the capabilities of chatty began? What are the limitations? As you can read, chat DBT cannot access Internet after 2021. Having said that the latest version of Chad DBT for you to access it if you're going on a paid version. But most of the time our questions are not around it. So let's use that. It may it very clearly says that it might generate I'm incorrect information is because they just want to be safe. And they also tell that please keep your data confidential. So do not give you a name band guard, either God or any of the important information in the chat too picky. Because what if it learns or it retains the data? It could be data enrichment. So I'm there on that DVD. I have created my account. What all can I do? I can write an email, I can ask anything I can code. As we will meet in a workshop, you will get an idea of how we're leveraging each of this. What is shown on the screen is a very, very basic questions and prompts which you can play around and understand. So this is the first part of creating your login ID on chat GPT, on opening a platform. As you're going to work with me on my boot camp. Sometimes we don't want to work on the website, but we want to work in our environment. And I will show you what is the next place where you need to create your login. So that is called as an Open AI logon ID. My personal preference would be that you create an account over here also, use the same email id or use a different e-mail id, the choices completely yours. So when you go to Open ai.com slash api, last time it was class chat GPD. The order to slash API. Api stands for application programming interface. You're going to invite chat GPT inside, or you are going to invite the algorithm inside your application and make it work for you. You can sign up over there or click on login if you have already signed up. So it will say, well combined give the e-mail ID and give the password and it will continue. The screen will look like this. It says, Welcome to OpenAI, Quickstart tutorials and some examples. You're free to go ahead and explore it. If you do not understand, that's perfectly fine. Do not worry. The feature that we're going to use over here is the playground. And plus we are going to use the API, which I will cover when we are meeting face to face. So just to get an idea, open, the ABA is not free. You will have to pay a price to pay czar in peanuts, right? So do not worry. The price is like 0.00 0441k tokens for an adult model, The Bobby's model, it is 0.3 zeros, 541k tokens. For QT it is 0.0, 02.4 Davin two to 0.02 different groups. So this is the pricing that I've taken as a screenshot from OpenAI platform. So don't worry, it's not at all expensive, it's less than one buys up. In the Indian rupees context. If you're generating images, you will have to pay them price because the content that you're creating will be yours. And you can go ahead and use it freely without having to worry about copyright issues. Important risk mitigation, right? Just like whenever you get on to the flight, we say that what to take care and what not to worry. It's important for me to tell you that what risk can happen when you're working with chatting. So there is a website that's just one of it you might find many other if you just want to chat Google and such antipathy and don't say open-air, you might land up in something like chat type and GPT dot ORT. This is a fraudulent side because if it is telling you zero out of ten messages lead, it's not the correct side because OpenAI does not have restrictions in terms of the number of request, you can please don't go to anything that does not have open AI in the URL. And it is open ai.com in the URL, whether you're going for the API screen or if you're going for you're chanting. So just be careful that you don't get entangled into a wrong website and never, ever give your personal information in the chat GPD conversation. If you are started playing with it before our training. Remember it. It's a wonderful tool, but it's important for you to be careful with your personal data. So don't share your band guard. Other cardio bank details are any of your personal information or dead. Even if you want to talk, use a hypothetical person's name and say that I was talking to radish rummage through creation, then go ahead and have your conversation. So I hope you are created two accounts, one on the chat GBD screen and other on the Open UI APIs. So see you soon. Bye-bye. 7. Let-us-play: As we understood the importance of this lesson, Let's get deeper. What does chance if VT, how does it work? We'll replace our jobs. How can I create an account and get interface? Tour of how I can use chat to let me take you to the place where chat diabetes more. Openai is a research company who has been working to create jobs. And as I told you earlier, the chat activity was originally released in 2018, plus the Washington one, not charge a bit. It was GPT, which was released in 2018 in 2022 numbers, the chat GPT-3 was released. By December 2022. It had already raised 1 million active users. And by January 2023, it had reached hundred million active users. So when you go to Open AI website to a new type of pump, the interface looks like this. And then you scroll down, it is talking about what are the various research they have done. The different products that they have. Chad DBT for double E2, customer stories, safety standards, and the pricing. What is the safety it takes care of and the company details. When you scroll down, you will find chat GTP and it says tried chanting. When I click on it, it takes me to a page that it actually should ask me for a login ID. But because I have already created my login ID, it has taken me to this page. I can go ahead and create a new chart. It shows me the examples of the chat. The chat DBT is divided into three parts. The right-side, bigger pain is the place where you are conversing or you are going to talk with the chat activity. On the left side, you see all the conversation history that you had. At the bottom, you see the settings. And if I want to delete the conversation, if I wanted to upgrade to a plus plan, I want to change the light more and some FAQs that if you have. So if you click on this, it will take you to the place where you can learn about the chat, GPT plaza release nodes, what are the general FAQs and so on. So now my first question was, how does it work? So I just come here and type the question so let's understand how does chatting. Yeah, so Chelsea bit is a large language model developed by OpenAI. It is trans, based on the transformer architecture, which is a type of a neural network designed to process sequential data even before I could complete reading. And you can see that it is continuously generating data, which is making sense. If I would have done the similar search on Google, it would have pulled up something like this. So let me do it again for you guys. So it takes me and it gives me multiple options to read and then find the most relevant answer. That is why you find that we go to Google to sort something. We get lost and our journey continues. When you come here, it clearly tells you what it is. So let me read it out for you. At a high level chat GPT works on taking the prompt or question from the user, generating a response based on the pattern it has learned from a large amount of text data. Specifically chat GPT use unsupervised learning model to analyze massive amounts of text and identify patterns in linewidth, including grammar, syntax, and context. As you saw that when I was typing, I was using the hyphen instead of the space. It knows that dimple. Always ask industry. That is why it did not confused because it gets customized to the user who is using it. When a user enters a prompt or question, the chat GPD process the input through multiple layers in a neural network computation using the attention mechanism and focus on the most relevant parts of the input. It will generate responses based on the pattern it has low. So it's not the same answer which we'll come to everyone. It is continuously learning, continuously take the feedback and it is important. Okay, so now let's go to the next question that we had. We understood that how it works. So we'll chat GPT, replace our jobs. So let's ask Chad, typically, when you read please. Jobs. As any AI model, I cannot replace humans in their job. My purpose is to assist and augment human capabilities, not to replace them. I can help with tasks such as answering, questioning, generating texts, and providing insight. But I cannot perform tasks that request physical abilities or human emotions. It is important to remember that AI is meant to work alongside human and not to replace it. It's such a beautiful response it has given him. That's the reality because chatty pretty by itself is also created by humans, the neural program. So I understand that it's an unsupervised learning algorithm, but the algorithm is written by something which we have pleated it that when you get a feedback work on it, you'll get a response. You have an option to give a thumbs up tearing that I liked your response. So it seems that can you provide me better understanding that what did you like? I would say it is easy to understand. So I get this feedback. So it will take that as a feedback. Right now. I can also play games with Changi because I can see it can give me a really, I'm always hungry. I must always be fed the finger I touch. I will assume don't read, Who am I? Can you solve it? Now? I don't know the answer to this, so I will see, please go wide. Now answer the riddle is fire. Fire is always hungry and it needs to be fed with fuel, wood, paper, or other material to keep it burning. If you touch a flame with your finger, you will get burnt and your fingers will turn red. It's so beautiful when you are talking to the chat GTP, it is giving you a response like a human, right? The only limitation which I see at this point of time is that it cannot do audio conversation and it cannot do a pictorial quantization. And I think in a few months we should be having that version as well. But as the name suggests, it is Chad DVD. So they might want to enhance the chat version of the text version of it because it kept these and we will continue more in the next chapter. 8. ChatGPT Get ideas for theme parties: Let's use chat GPD for some ideas of theme parks. It's my daughter's birthday party and I want to plan something. So I would say, give me some ideas or themes. Body or mind. Is. Now if I did something like this, it is going to give me a generic arms and let it dry. I'm going to tell you how to input your prompts. So it's thinking fairy tale party and greet your daughter and her guests to come dressed up as per their favorite fairy tale character. Decorate the party area with fairy lights, colored balloons, and fairy tale Team Declaration. Isn't it wonderful? Princes party because he's a Dr. they're telling their local transform your body part into Royal Ballroom and have the guest ran their best princess dress. Wonderful. Now I will just drop it. And I will say that my daughter loves Barbie. But I am in I constrained to spend 5,000. I am not well defined. Now. Give me some interesting ideas. So now, oh great Bobby theme party sounds like a great fun for your daughter's. Each row divided. Bobby declaration, use pink and purple streamers and bundles to create Barbie the integration Barbie games. You can organize some Barbie thing. You can do some tweets, which are being pink and purple cupcakes. I loved the idea of the cupcakes, but I don't want to go out and buy. It says you can create DIY, but Bobby K, by decorating a plain Kate meant Barbie dolls and Barbie accessories. Remember, you don't have to spend a lot of money. This is a beautiful to create, a fun and memorable party here, Awesome. With some creativity and beer, I buy skills, you can do it. Now I will say, Can you help me? Kids at home? And I want the key to beat. I'm adding some instructions. Here's the Eglance cake recipe. One cup of all-purpose flour or sugar, baking powder, baking soda, salt. Now you'll see that it's helping me plan out the antibiotic party right now, heat the old one and do it. It's giving me the whole process. So you can use this for creating it. Now, I'm going to get kids at home so I have to keep them busy. So I want to let it complete the recipe. You should have 12 to 15 cupcakes and you can add food coloring and batter made the cupcakes pink and purple grid. Can you suggest some gifts? That little goods. Lady. And I can organize efforts. So now it's going to tell me how I can organize the team party and what type of Gibbs can I pay the least amount of findable on the barbie? Oh, that's a great idea. I never thought of. So clearly large poster of Barbie doll without a bow on her head and blindfold each player and ask them to go and put that wonderful. Bobby got it. Right down the various Barbie related phrases and actions on a small piece of paper and put them in a bowl. That's also a good idea. But we fashioned challenge, divide the girls into teams and give them back. So random clothing items and accessories. Oh, that's also a beautiful idea. So you can see the challenge if it is helping the planet and n-type birthday party for my daughter with all these things. Now, what if I was not planning a party for my daughter? Instead, I was planning a party for my phone. I would say. My best friend is 40 years old and we know each other or 20 years now. So it's not a kid's party, but somebody you can waterboarded plenty years of friendship party theme and you can see celebrated as a 20th and chip team photo display. Create a photo display of your favorite movements together in the last 20 years. Memory lane quiz, graded quiz of questions about the authentic over the last 20 years in capsule. But it gets bringer item that represents a special memory and movement in that friendship. Are MQ amazed at the quality of thoughts it as giving? If I would have gone to Google, I would have not got sadness. I would have to do a lot of, lot of research here. It becomes very easy for me too. Go ahead and do it. So it has given me a beautiful five ideas. I can pick that and do it. I even like the BMI set up a DIY crafts teach and lead guests can make friendship bracelets and picture frame and graphs to celebrate. So making them feel like a kid again. So spending time with their loud words and celebrating the memories is a good idea. Interesting. You will find that they really, really appreciate the venue. Complimented. I'm glad you liked ideas. Our fun plan in the party, and I'm sure your best friend will appreciate therefore to put in creating a special set of features of your 20 years. Isn't it amazing that you can use charts EBT, not only for your productivity, but even making your personal life very interesting. So let's get going. 9. ChatGPT Playtime Word Guessing Game: Let's take the next lesson. The chat DBT playtime for guessing game. What if I'm getting bored and I don't want to play any video games or neither. Am I interested in watching Netflix. I can use my time with Chad TBD to become intelligent and also play some fun for games, like the word guessing you. Let's get started. I'm going to judge you if you do screen, see how well I can have conversation with transitivity because I'm feeling good. Hello as an airline which I do not have any feelings. But how can I assist you today? I wanted to play a guessing game. I can certainly help you. These are many board guessing games that you can play. Hangman one day and things of the world in his mind. And it goes off. 20 questions. One player thinks of a word, and Adam tries to guess the word by asking Wendy yes or no questions. I like that, but so I said I like the auction. Let you go and ask me questions. One of the other plane rise to ask. Let me see. I will let us play this game. Do I will think of the word and you asked me that question. I'm being very specific. So now I'm thinking of the word like charge if you do. Sure, I'm happy to play a game. I think of the words and I will start. Please think of the word. I start guessing and try to guess this. I'm ready. Let's start. Is it an object? Is it something that you have found off? And it's asking all the questions at one go. Is it a bread box? Is it a food item? Is it a form of technology? Either something is a two type of tool. Is it a natural resource? As I say, it's a form of technology. Use grid. Based on your answer, it seemed that the word you're thinking about is a form of technology. Here are some additional questions that can help you narrow down the possible easy to debug a computer screen. Is it a mobile device? It to the entertainment? Is it useful work and productivity? Is it Internet connectivity or is it home automation? So I would say it is for productivity. So I say it's six. Because I thought of charging. Based on the answer, it seems that you're thinking of a form of technology that is used in work is broken. Is it a software and hardware other tool that is used for writing and editing? Computer isn't a specific region specific industry equipment. Does it involve using the Internet? Yes. It uses a software. I'm saying yes, it's users and a negative type of software. Thanks for that additional information based on this. So what happens is, isn't a web browser-based social media platform. So I say it's true. It's true. Yeah. Seems that you're thinking of a web browser. Is that correct? No. So what happens is, I'm confused. It asked many questions. I did not ask. Uncertain. Is it used for accessing and manipulating data? Is it a database software? Is it a file? User productivity? So it's thinking about something, but it's not thinking about itself. But it keeps me a dividend. Because now by reading these questions, I'm thinking, what are the different things that happening in the world of software? And IT, isn't it wonderful to be using this in your free time to upskill yourself and get some general knowledge. Wonderful. Let's get scattered. Continue in the next chapter. 10. ChatGPT Playtime Poem: Let's keep all gets busy. Or sometimes we want to influence somebody, but I think that's it. Good boy. But if you are somebody like me, you're not good with writing poems than chatty be they can help you. So let's go to our chat deputy screen. I'm healed at the chat DVD screen. And I want to write a beautiful poem for my daughter. I see, right? Beautiful foil for my seven year old daughter. Let's see what activity has for us of all the treasures in this work. There is nothing quite like you. My precious little angels, so sweet and kind and true. It hasn't done the rhyming so beautifully. My eyes your eyes are like ocean, so deep and full of grease. Use mine is like the sunshine that's light up every piece. You can feel my heart with funders. And that's each and every day. And through your boy growing quickly, you're always buy my array. Now you can see that it is written the way we would for Ryan. And you will find that at this point it might not be the same if you are writing the same question on your chat to charge you for degenerate new poems for office. So it is customized for each one of us. So it's not that if I write, you will also get the same answer. Let's say I want to write a poem for my school. And a pretty **** on graduation. I'm giving some more context. I'm adding a line and say, I really love. Let's see what antibody has my dearest daughter. On this graduation day. I can hardly believe how time has flown by. It seemed just yesterday you will in China, but now you're grown and I'm filled with pride. You have you have worked so hard to law and sleepless nights, pushing through each challenge everyday. I know the road has not been easy, but you have always been strong and steady. I have watched with amusement as you have goes into the beautiful woman that you are shown. Now, as you graduate and embark on a new path, I know you will continue to shine and make us love. So hold your head up high. My precious one. Know that you have only just begun. I love you more than words can say. And I have always been here to support you every day. So take this moment to celebrate and bask in the glory of all your accomplishments. So you can see the language of the poem for a seven year old versus a 22-year has been changed according to them. You can also write your poems and uploaded as your project. As part of this, I will continue with more things in the next lesson. 11. ChatGPT Playtime Riddle: Let's explore more capabilities of charity, beauty. If you have kids at home, then it's important for us to keep them busy with some activities. So let me take you to the screen and let's see how Chad UP they can help us. It's the red unpaid playtime. How can I use chat DBT to play riddles with me? We know that gets laboratories and v as bands are now out of freedoms. We have already bought many books, which make us keep them busy, but still they want something new every day. So let's go to our chart ZPD screen. You need to go to openai.com. When you click on Try to activity. If you haven't already made Chad dot openly i.com as your favorite, then you would be seeing the screen. I'm going to ask correct, wrong with can help me ask the right questions. Please. Email me or for my seven year old daughter. Which makes API to start thinking. And here it says, here's an adult for your seven year old daughter. What's the harm that doesn't beat? The answer is the antigen. For this. She enjoys it. I do not like the reader. So I'm going to ask, can you give me anymore? It's important that I keep asking questions till the time I find the right answer. Because prompt engineering is very important. So here is another riddle for your daughter. I'm taken for a Mac. I'm taken from a mine and shut up on wooden case from which I never released. And yet I've used by almost every person who am I the pencil lead. So the pencil lead is inside the wooden case. So beautiful. I see, I like it. Can you give me more? So it's important for me to have a conversation with my purple rigorous so that I can keep my child busy. I'm not alive, but I can go. I don't have lungs, but I need air. I don't have a mouth, but I need water to survive. Well, am I dancer is fine. It's beautiful, right? So you can use chat DPD to create some fun questions like this to keep your kids busy with intelligent agents. Let's continue in the next chapter. 12. ChatGPT Clean up the email and mail merge: Hello friends. We have been hearing about chat GPP a lot. It's a new AI tool which has changed the way the world is working. Most of us feel that CBT is about going to the browser option of Jan DVD and working, asking questions, working over there, getting the answer. Can I take the capability of tragic to be using some small features to make my life much more easier. Let's assume that you are a marketing lead seamlessly who has a job of sending out hundreds of emails on a daily basis. And you have to keep track that how many emails have bounced because the e-mail ID was not correct. How many emails have been opened and read successfully? How many emails? They have clicked on the attachment or the link that is given in the email that you have set. If all of this could be tracked easily on a Google Sheet, my life would be a blessing. Today. In the next ten to 15 min, I'm going to teach you the capability of using GPT to solve this type of business problem. So stay tuned. Let me share my screen and at any point of time you get stuck, don't feel free. Hesitate to ask questions in the chat section. So I have opened a simple Google sheet. How do I open a Google Sheet? I log into my e-mail ID. I can see nine dots on the right top site. I scroll down and I have sheets. When I open up Google Sheets, it opens up like this. I can open up a blank sheet and I'm ready to one. So I have my sheet opened over here. I would need to treat things before I can get started with my work. Number one is that I need to have a login ID on platform dot open-air. So if you go, you can see that I have logged onto platform dot open ai.com with my login id. The screen looks something like this. I'm not taking you to the charging the screen. Chat DBT is one of the products by open-air. I'm taking you to direct platform, which is built. So what I'm going to do, I'm on Open AI platform not opening i.com. I would request you to open your account over here. It's absolutely free. Once you open your account, you will see a menu like this. Over here, I have an option called as view API. Epa, as many of you who are programmers might be aware that it's an application program interface. It means I'm going to use this key from my Google Sheet to go and unlock the potential of the Changi VD or the products that openly I have to offer to me. Here I'm opening eyes website. I'm at my API key, and it's simple for me to create a secret key. So I click over here and I'm going to say for my e-mail track, it's optional to give those keys. It's good to give the key because you might be generating multiple keys. Some of the keys you might want to discontinue way forward, so you can go ahead and revoke. The key comes up, I'm copying it, don't worry, I'm going to delete this key after the recording because I don't want others to use my right. Similarly, I will advise you also to keep your secret key as a secret. Right? So let's come back. I have my key, which I have taken for now. I'm just going to paste it over here in my Excel sheet so that I can add my Google Sheet to keep a track of it. Because once the key is created, there is no way that I will be able to read that key again. I can only get changed the name, but I will not be able to read the key again as a reference. I went ahead and copied Mike. How do I believe my key? I just have this reward key button, which is a small bin icon on the right side. If I click on it and say reward, this key is deleted. And even if I tried to use a key, it will not work. Very good. So you are with me. I hope you have opened your account with platform dot open-air dot com. And you are in the view API keys and you have gendered expectations. Once more important point. Most of the time when you open a new account with platform dot OpenAI, they generally give you a free credit anywhere between $5 to $18, right? So sometimes they don't give you this credit. If they don't give you this credit, this thing is very simple. Just go to the billing section and say set up a paid account. You can set up a paid account as small as $2, which is like including GST will be less than 200 Indian rupees. So coming back. I have my $5 as a credit and I can see that I can use this credit till 1st, August 23. I have my API keys where I can generate my keys. So I have set from the platform for now, because I've deleted this T, I'm going to remove it from here. I need to go to the extensions, go to Add-ons, and click on Get Add-ons. This, I'm doing it from Google Sheet. So once I go there, I need to click two types of add-ons that I would need. First one is DPD, four sheets and dogs. The first one. Right. So as soon as I said DVD for sheets and dogs, you can see an option like this which comes up to you. As of today, we have 2 million downloads which are available. I have already installed, so that is why my option is coming as uninstalled. The second add-on, I will explain you in a bit. So once I have my extension add-on attached to what? You will see that GPD for sheets come here. I can click on Launch and an even function to check if my API key is active or it's not working. If you're doing it for the first time, you will have to set the key. As I deleted my key, it is telling that your API key is not working. I guess I should go ahead and change my. So how do I do it? I come here, I can read my new key. I copy this same done, come back, come here and paste it. Icc my API key. This key will act as a bridge between your Google Sheet and the OpenAI platform exceeds your API key has been saved. Now we are ready to work and let's get started. So the first problem when we work with the Sates Department in the sales department is that we receive multiple emails. The email IDs that we receive as part of forms are not in a easy to use for it. I'm just showing you up some of the emails that I have received from my from my survey that I have launched. Somebody has returned, we know dot n at Gmail and they forgot to give a dot. Somebody has given their names. And they forgot at the door. Somebody gave the name and in angular brackets the e-mail ID. Now, I will have to sit and manually change this to a easy to use humility. I, what I did is I have given a sample of three email IDs which I've corrected manually. I copied these three. I pasted it here. And I said what is the correct way? So I just let me do it. So it is going to be.com, Swami or india.com. And I don't need the name because I'm only interested in the e-mail ID. Dark color. So I have my Google sheet. These are the wrong these are the raw formats in which I have received my email. This is an easy to use format because as soon as I hover my mouse over it, you can see that I'm able to save it as my contact details to my Gmail account. I can go ahead and save it as a contact. I can send any way. I can set up a Google Meet. And I say can set up an event. I want it in this form. So the first task I'm going to do is clean up my e-mail IDs that everything. I'm going to use a simple formula for less GPT fill. It says, give me the examples for me to start working on it. So I have given three rows of examples, three bad ideas, and three keratitis. Now I am telling, Please do this work for the remaining open. Sorry, I clicked before I finished. I need to do select all the main ideas from. The first input is the sample, which is wrong versus carrot. And the second input is what I want to correct. I just press my Enter key. It says loading. Do not worry about this error message because it's still loading my data. I just have to give it a few seconds and it will be done. Can you see how beautifully it has converted all the email IDs into easy-to-use format. And you can see as soon as I click on it, it is activating my main writing facility, right? Even though I had not trained it on something like point Shiva as Yahoo coin, it was able to understand that it is yahoo dot c or I don't have to write multiple formulas with left, right, and center. As we have in cheats and exit. I just said, I taught it how to read and it has character for me. Now, the next step is to write emails to these people. Mail merge is a very easy option that we have learned for decades now, where I have a standard template and I go that can you please replace the firstname with this name and so on. But the overall structure of e-mail remains same. I'm going to teach you that. I want to write a customized e-mail for all of them. So in case two of the peoples are friends, they might still feel that I have personally sat down and written the ene. So what am I going to do? Let me just take you to the next sheet. So I'm going to tell my sheet certain instructions. I'm saying I'm going to paste the prompt or the question in column a, and I'm going to give you details in column B. So the tasks that I have is to write an email. The style that I want to use is a creative style. I can use any style that I like. I can see professional resume, a cover letter, proposal, and so on. The bone that I want to use a professional tone, I am free to use a friendly tone, also. Great. But because I'm going to sell a product, it is better to use a professional goals. I will change and actually put option. The topic on which I want to write an email is called cell. The program by name secular your carrier, and be proficient in by using AI. So are you clear with me till here? I want to write a creative email in a friendly tone on the topic. To sell a program. I want to start with the word hi and use the FirstName. This first name is, well, what is m giving over here? I'm also giving certain more further constraint telling that relate that job title to the topic. So job title is and relate it to the topic that we are selling. This is the topic. Give examples of their hobby and how it will help them. So I have reading, traveling, painting, dancing as the hobby. And I want it to be related to it. And I'm telling sign it off as dimples hungry, I'm giving install module link for them to go ahead and click on the link to see what happens. So I'm just giving this. Now what I did. I have few email IDs. I just cleaned up the email IDs over the array. I'm going to copy all of it. Just a second. I want to send emails to all these people from copying this coming back or you're pasting values. Okay? Because if there was a formula, it was pasting it as a Soviet ready. Okay. And I can do one more piece. I can fill if I already had the details of first name, job title, and so on. I can fill that up. If I want to pull out the first name from the e-mail ID, I just need to do the same thing like Rami. I'm just delete teaching it on three and I'll say now, please fill it up. I'm using the same concept. If it isn't. Please understand these three are the examples. And from here, I want to pull out their phone. Yes, Given 2 min, 2 min, I should say a few seconds and it will pull out the names and voted on it. Yet see, it pulled up Sedona to revalidate. It was able to identify that from your work in that I can so I have my e-mail IDs, I have my first name, I have the job title for B and company. It's not compulsory for you to follow the same format. But these are few things which I'm using over there and I want to use that. So now it's time for me to write emails. I don't want a standard email for all of it. I want a customized e-mail for each one of them. So then there is a very beautiful formula which I get to DBT. So I'm going to first create a prompt. So I'm going to say GPT, create prompt. I'm going to give arguments. What are the arguments? If you see, I have some prompts and some details, so I'm just going to copy, select all of this because it is going to be fixed. I'm just pressing Function F4, inDollar be dollar ten. Now, there are another set of arguments. We can e-mail ID, first name, job title, hobby and company. I'm going to pick that again. I'm fixing the location by pressing Function F4. I close this bracket and give Endo. So you will see that squiggly, it has generated a prompt for me. Let's read the prompt. What it has generated. So what is happening? It's telling prompt is the task is email style is creative, tone is friendly. Topic is send the code start with hello, hi friends, and relate the topic to the job title, give the examples of their hobby and so on. So the prompt is created. I want to now send this prompt to toe opening. So I'm going to close this as part of my TP. So I'm going to say GPT, that is the prompt comma. I have to use the values because for every p, The value for every email, the values are different. So this time I'm going to sell, selling the values. Um, remember, I'm not pressing function f for this time. I'm just going to press Enter and see the magic. Let's just give a few seconds. Yes. So see, I have my e-mail that is ready. Let's read the email that it has generated for the first-person. Let me align my text so that I can read it properly for you. So highway node, as a CEO, you understand the importance of staying ahead of the curve and staying competitive in the market. With the rise of artificial intelligence, it's more important than ever to stay ahead of the latest tools. It has very beautifully crafted. It, it has given a email drafted. Now, I said please sign it off as dimples and we start with the first name, relate the job title to the topic. Relate his hobby. So let's see if it related, right. Plus you can apply the skills you learn to your reading hobby, allowing you to explore the new possibilities to gain deeper understanding of the material because the person loves reading. Now, I'm just going to do one thing. I'm just going to drag this down all the way for all the emails that I have. It will just take few seconds and I will have beautifully crafted email customized for each of the e-mail ID that I have, each of the record that I have. It's just beyond male mode. So you can see it has generated for the second one. Let's read it once, right? Yes. You can see us as a freelance trainer. You understand the importance high energy, Anheuser Vernon, and so on. So when you see all these e-mails, you can easily realize that as an RPA export, as a digital marketing, as a trainer, as a manager is taking it up and the sentence formation is also different. You understand the importance of staying ahead. Here. You know how important it is for you to stay ahead. So there is slight bit of customization that has happened. There is another way for me to increase this customization by changing the temperature. So if I come here and then see temperature is equal to 0.8, you will find that it's just getting a little more creative. Let me just pull this out. Give it a few seconds and we'll be ready. Yes, see, you have the ability to shape the future of your company. And artificial intelligence can provide powerful tools to your journey. Here it says, you are looking for ways to boost your carriers. No doubt familiar. So by default, the temperature is set at 0.5. I just asked it to get more creative. Now, I don't want a standard subject line for it. So I want a customized subject line cord. So I'm going to say equal to GPD. The prompt is create 0. N. E means subject line based on I close the brackets, give a comma the value was I close it. You'll see it comes up with a type to unlock the potential to secretary. Okay. I'm loading it down to Curia. Curia with the idols because I gave a restriction that I want it in ten words. It is trying to do that. If I just say Create, create to email subject line based on this, I find that it is going to change the way it is working. Right? So yes, I love the potential to secure your carrier v-node. Stop here. I'm just going to pull this down. For all the emails that I have done. It is telling loading, but you will find that soon. It has everything ready. Yes, see, unlock your potential. Fourier. Join us here. Some places it gave the name, some places it was just creative in writing it down. So now I am ready with my e-mails. I'm ready with my subject line. What's the next task? The next task is for me to send this email to my clients and the email id or present over here in column D. So let's get started. For ease, I'm just reducing the size of the row to make it a little easier for me to read. So I would need one more small extension that is available. So I go to Extensions, go to Add-ons and say Get add-ons. Once I'm here, I will just go and say, yep, that is yet another mail merge for Gmail. So I'm just pulling that up. So I'm getting this one. So I'm taking yet another mail merge. I'm going to install it. It's telling that it requests permission in order to get started PC, just go ahead and say Continue. I'm giving the bomb mission. And just go down and say aloud. And then I'm also going to tell you how to remove the permissions once your job is done, right? So it hasn't started Next and I'm going to close this. It's loading my young. Just give it a few seconds and you will see the magic happen. This add-on modes template created a draft in the Gmail and please create the dropped in the middle and try again. Okay. So I need to create a new draft in my e-mail, which I keep it over there, which I want to sell now. So let me come and say compose. I have my compose. It needs certain instructions. The instruction is, I want the email body we pasted over here. So I'm going to paste it over here. And I want the subject line to be pasted in the subject. So the very simple dropped that I've created, subject line should come here and e-mail body should come here. Nothing else. I'm going to close this in my drafts. I have only one drug because subject line in email body, because the e-mail IDs are going to be picked up from my Google Sheet. So let's get started. So I'm here. I see yet another mail merge and start the meeting much. Yes. So it sinks. Start the main mode because I have only one dropped. It will pick up the right drop. Otherwise, I will just show you how. So, yes. So sender's name is Emily template. I have only one template, subject line, one Gabi. This was created. If you have multiple templates, you can go ahead and select the right one from the drop-down. Now, I have clicked or tick mark, grab the email is opened. Did it get clicked or did it get bounced? Right? I can either schedule my email, I can send a test email, I can send all the 29 min. Right? So what I'm going to do is I'm going to click on Send 29 e-mails and just wait for few seconds. So it's shooting out the emails. Now just to let you know if somebody has given me a wrong email ID or whatever has happened, I will get to know over here because it sounds very familiar. All of them are gmail.com, yahoo.com or Tata steel.com and so on. E means you can see how it is progressing. All emails have been sent. It says 13 emails were sent. Why? Because some of the emails did not have the recipients. Okay. So I had only 30 min. Beautiful. All of them have been delivered. Know, it's getting replaced. See, I have immediately got a pounds which all e-mails where a failure. So they tried to deliver out of my 13 emails. Eight emails have been bounced. Perfect. So I don't have to go to my e-mail ID and check which emails with them. Now let's do a simple magic. I have certain, so I will be able to add some here from the attacker, or I have a new column that is created. Let me do a simple thing. I haven't opened up the email this just check for this. Let me open up my email on my mail, on my mobile. And you will find that it is going to tell me that the email is opened. I'm still just getting updated. I just opened my email. Okay. Yes. Just get updated. I can open the tracking report. See can you see it says one email has been opened and I responded back. So it says you have also got a respond back on to see which one got the response. Can you see it has automatically drawn to green because I opened the e-mail from my mobile to show you that how it works. So it becomes so easy for you to track and do things that were there. I haven't clicked on the link that was sent as part of the image, and hence, it is not saying clicked, I just responded back, thank you. And you will be easily able to track that. How are you progressing? I asked you to send out e-mails, but I have only given 13 e-mail ID. So it has sent all of it. It stopped or what you are saying there are no receive notice it notice it BMT e-mail IDs, which is cut. To learn more stuff like this. Stay tuned with us. Thank you. Bye-bye. 13. ChatGPT Starter Code Prompt: Let's cope with commands. Here is my prom that is very clear, but it's not very specific. Read the data from the CSV file and filter out all the records where the age is less than 20. I don't specify the technology. I don't specify how the file has to be read. Now, Chat GPT has gone either way, but it chooses Pandas for this. It installs Pandas, generates the sample CSV file with records in a certain format, and then uses Pandas library to read the data and perform the filtering. But maybe this is not what I wanted to use Pandas for or I did not want to use Pandas at all. In that case, I have to be more specific in my prompt. Can you please use the CSV reader for this this is also a follow up question with CharPT to correctly interpret this. It sets up the CSV file and then uses the CSV reader to perform the same reading and filtering operations. While generating code, I have empirically found that it is far better to be a specific concrete set of instruction. Here, I have just used the triple codes, and then I have three different steps within the Triple codes. Step one, create a list of employees who belong to the department Step two, create a list of department names and the IDs. Step three, combine them to get the result of five employees with their departments. These are precise and clear. These clear instructions, you will likely get good granular step by step code. Now, I haven't specified a programming language, and hence CHARGPT automatically generates data using Python because my previous questions were using Python. And it feels that I want to continue my conversation. It is also quite possible that whatever code or comments it generates could be in SQL. My instructions were clear, but I haven't specified the technology, which is why Chat GPT is free to choose. It picked Python because most of the past interaction here in this conversation, where about Python. If you were doing the code using Java, it might have used that. Chat bots are powerful using natural language models. And when we have discovered that, we are going to use them. They have a natural understanding of code. Now you have generated codes using English statements. That's it. Just the prom that describes what you want. You have seen that you can also generate code by specifying your prompts using the code commands. You can actually get even more technical with Char GVT and generate codes by providing some shorter or a starter code. If the starter code is meaningful, generative AI models powering these chatbots will basically just try to complete what you have started. Remember, for this, your starting code should not be ambiguous. This code should be clear. Define what are we looking at. If possible, include the comments. Now, I must tell you that I have been so excited when I first encountered this feature, and I made sure that it worked on hat GPT. And then I checked Bard as well. Just to make sure Bad understands the code exactly the same way, we will find that it does. Let me show you what I mean with an example. Let's say you want had GPT to write a code for some function. It's a straightforward function that Chat GPT would have understood with the descriptions that I have given, then it would go ahead and create the function with a meaningful name and then hat GPT gives some starter code. Let's say the function signature. And from the function signature, hat GPT is able to tell which language you want your function to be coded in. Here, it's pretty clear that I'm coding using Python. Well, it's all the prom that chat GPT needs to complete the definition of this function. It should be pretty clear that you are looking for this function. Let's say I want to build a function e prime number. It's pretty clear instruction to the chat GPT, and the fact that I'm going to use def keyword, I have specified the colon at the end of the function signature. This indicates to chat GPT that the code should be generated using Python as a language. Now, let's see how Chat GPT has to say. You can see that it has figured out my intentions perfectly. It even explained things to me. It seems like you want the function e prime number to check whether a number is prime or not. Notice, in the function definition of the generated code, I have taken a single input argument. Now, when I have defined my function signature as part of my prom, I did not specify the input argument, but Cha GPT still added it because it made sense to it. So it's pretty sensible about coding this way. You can look at the rest of the code and see that the code is actually correct for the prime number. It is able to check whether the number sent in the parameter is prime or not. I really appreciate that these chat booards are very helpful for people like us when we want to create functions and do exactly what I want the code to do. Now, let's see if Bard can do this. I have switched over the tab. Here I have Bad open. I'm going to give Bad the same starter code prompt. Def is prime number. I don't specify the language. Let's go ahead and submit this. You can see that Bad has also recognized what I'm trying to do, and it has generated the right kind of code. The code generated here is almost exactly the same code that the chat GPT had generated. You can also see that the explanation for the code here is at the bottom as well. So clearly Cha GPT can do it and Bard can also do it well. Now, let's go back and give Cha GPT some more starter codes. This time, I will ask Cha GPT to validate an email address. Notice that the structure of my startup prompt code, I have void, and then I have validate email address and the string email in the input argument. And then I have opened the curly braces. Clearly, this is a code in Java. The way the method is defined and the structure of the function. Everything points that. Now let's see if Chat GPT has to say. My starter code prompt was very clear. My intent was very clear. Now Java code has been generated. If you look at the function signature in my prom, we can see that the return value is void. Meaning, after validating the email address, the code must not return through or false. It must either throw an exception or do something else. Let's see how hatGPT has done this because it's Java. Chat JBT has placed the function in its coded up and inside the class email validator. The class has main with few invocations as well. Notice that Chat JPT has not tried to return anything from the function. It is just simply printing it out on the screen. Whether the email that I have passed as a string, is it valid or not. Then, of course, we have the explanation for the code at the bottom. The code uses regular expressions and the explanation included in it. Isn't this smart and helpful? 14. Master ChatGPT Open ended n Multi part n Advance prompts : Mastering hat GPT prompt. In this section, we are going to cover about how to work in the world of prompts. We will dive deeper into the world of prompts and explore more advanced techniques that will help chat GPT generate more robust and relevant responses to the user queries. We will describe advanced techniques for writing prompts for chat GPT. Evaluate the effectiveness of chain of thoughts prompting, outline the concept of zero shot prompting, and summarize the concept for self consistency. We will describe the concept of applied programming and describe how chat GPT can solve multiple choice exam questions. Well, we will outline the steps to build a chat bod that is connected to a knowledge base, learn how to minimize common chat GPT reliability issues, such as hallucination, flawed explanation and biases. Identify potential biases in the chat GPTs response. And we will discuss the prompt debasing techniques, the term prompt hacking and gel breaking in relationship to the chat GPT. In this topic, we will value the contextual information of your prompts. Contextual prompts include aspects like background information, historical context, or specific details to provide a more comprehensive settings. Adding context to a prompt can help direct the generated response towards a specific topic or an angle. Let's look at some examples of what a contextual prom may look like. Example one, the roaring 20s was a time of significant change in the American culture. How did these changes affect the role of women in the society? This prom provides historical context by referencing the roaring 20s. And we're asking about a specific topic, the role of women in the society. Now, let's understand the example too. In the hit TV show the Game of Thrones, the character Joan Snow is known for saying winter is coming. How does this phrase relate to the overall theme of the show? In this prompt, providing the contextual information about the popular TV show and ask about a specific aspect, the relationship between the character's catchphrase and the overall theme of the show. Let's understand the third example, the COVID 19 pandemic has had a significant impact on the global economy. How has it affected the technology industry specifically? This prom provide contextual around the current events, ask about their effect on a particular industry and the impact of COVID 19 on the technology sector. Now, let's understand about open ended prompts. Ask broad questions that allowed for a range of responses and encourages creativity by char gebty. These types of prom can result in a variety of responses, making them ideal for generating ideas or exploring different perspectives. Let's discuss some of the examples of open ended prompts. What are the potential benefits and drawbacks of using artificial intelligence in healthcare? Here we are asking a very broad question that allows a range of responses, facilities creativity by asking about the potential benefits and drawback of AI in healthcare. Let's look at one more example. What impact will climate change have on the world in the next 50 years? This is also an example of an open ended question that encourages a wide range of possible responses on the topic, impact of climate change over the next 50 years on the world. Let's dive into one more example to understand open ended prompts. How will technology change the way we work in the future? This question is inviting a range of possible responses that explores how technology will change the way we work in the future. Hence, you are understanding that a prompt can be an open ended prompt. Multiple part prompts can ask several questions or provide multiple avenues for exploration to encourage more in depth responses. These types of prompts can help generate more nuances, responses and encourages reader to consider multiple perspectives on a given topic. Let's explore a few examples of multipart prompts. Prompt example one. What are the causes and the consequences of the decline in the bee population? How it might decline the impact on our food supply and the environment as a whole. What can be done to address this issue? This prompt consists of multiple parts, multiple questions that addresses different aspects of the topic, the decline in the bee population, and it is also encouraging hat GPT to consider different angles and explore different potential solutions to the problem. M. More example is, what are the benefits and drawbacks of remote work for employees and employers? How might remote work impact the work life balance of employees, and what are the effect for the productivity of employees for the company? This prompt provides multiple scenarios from employee's point of view and employer's point of view. The question that explores the facet of remote working. This prompt will give you different responses as shown on the screen. How has social media changed the way we communicate? What are the potential benefits and drawbacks of this change? How can we use social media more responsibly and effectively? This prompt consists of multiple questions that addresses different aspects of the topic, social media impact on communication. It encourages readers to consider both the positive and the negative of these ambiguous technology. The first question prompts the reader to explore how social media can change the way we interact with each other. The second question asks to consider the potential benefits and drawbacks of this change. Is it increasing connectivity or heightening polarization? Finally, the third question encourages us to think about how to use social media more responsibly and effectively, given the potential consequences of unwise or excessive use, by breaking down the topic into multiple questions, this multip prompt encourages deeper thoughts about the various implications of social media impact on communication. Advanced prompts are a helpful tool for encouraging user experiences in various fields and applications. We will explore examples of Advanced prompt and learn how they are being used in different industries. Chatbots have become increasingly popular across various industries, and advanced prompts are an essential components of their success. For example, personal finance chat boods that are about spending habits, set financial goals and provide advice and insights on how to use advanced prompts to offer more personalized financial advice. This is based on the user spending pattern. Health and Wellness chat boods are Health and Wellness chat booards can use advance prompts to provide tailored health advices. It tracks users daily routines. Retail chat booards can ask about style preferences and can recommend products. They offer personalized fashion advice based on users browsing and purchase history. Advanced chat GPT prompts can also be used in education of prompt that critical thinking, creativity among the students. A science education platform can use conceptual prompts to provide background information and encourage exploration and experimentation. A language learning platform should encourage use of open ended prompts to promote practice and creativity, such as writing or speaking. These prompts allow for a variety of responses. A history education platform should use multip prompts to encourage in depth analysis and discussions about the historical events. Marketing campaigns can benefit greatly by using the concept of advanced prompts, as they are providing generalized and personalized, engaging experiences for the customers. By tailoring the content of each individual preferences, marketers can create more meaningful connections with their target audience. One example of using Advanced prompt in marketing is through interactive quizzes. These quizzes can use open ended prompts to offer personalized recommendation for products or services based on the customer's history. This not only personalizes the experience of the customer, but it also helps the marketer gather valuable insights into their target audience. Social media campaigns can also use advanced prompts to encourage user generated content and discuss multip prompts can be used to ask users to share their experiences or opinions on a particular topic. Creating a sense of community and engagement among the followers. Another example in marketing is through email campaigns. By using contextual prompts, marketers can offer personalized recommendations or offers based on customers interest and behavior. This approach can lead to higher engagement and higher conversion rate for the campaign. Advanced prompt can significantly increase the effectiveness of marketing campaigns, whether in interactive quizzes, whether in interactive quizzes, email campaigns or social media initiatives. There are many ways to incorporate advanced prompts into marketing strategies. Advance prompt can also be powerful tool for inspiring creativity and providing new ideas for creative writing projects. By offering a starting point or a constraint for a story or a poem, writers can push beyond their usual boundaries. Explore new territories of writing. Thank you. I will see you in the next video. 15. Master ChatGPT writing prompt generators n Scenario Prompts: One example in creative writing is through writing prompt generators. These generators use open ended prompts to provide the starting point for a story, or maybe a poem, which can help writer overcome the writer's block, and it sparks new ideas. By offering a range of different prompts, writers can explore various themes and styles and genres to find inspiration in their projects. Another way to use Advanced prompt in creative writing is through writing challenges. These challenges use contextual prompts to provide specific details or constraints that can inspire new ideas or approaches. For example, a writing challenge might provide a prompt that requires the writer to incorporate a specific object or a setting into their stories. This can help push them outside their comfort zone and explore new possibilities. Finally, writing exercises can also benefit from the use of advanced prompts. Multiple part prompts can be used to encourage exploration and experimentation with different genres or different writing styles. For instance, a multipart prompt might ask the writer to create a story in a particular genre, but with a specific twist or constraints. This can help expand the writer's way and encourages new approaches in their craft. Another popular use for Advanced prompt is in data science, where you can improve the data quality, quantity, and diversity by generating synthetic data. Creating diverse variation of existing dataset and extracting features from raw data. This will provide useful in three main applications. Data generation for machine learning models, GPT based language models can generate synthetic data to improve data quality, quantity, and diversity. This is particularly valuable when dealing with limited data availability or when working with sensitive data that cannot be shared. Next is NLP, natural language processing. NLP task uses the concept of advance prompt, such as sentiment analysis, entity recognizon and topic modeling. These helps extract valuable insights from the text data, which can be applied in areas like customer feedback, analysis, and social media monitoring. Predictive modeling, GPT based language models can be used to develop predictive models for classification, gorion and time series analysis. These models help organization make accurate predictions and informed decisions. Finally, advance prompt can be used in programming to improve code quality and efficiency. By generating high quality code snippets and scripts, providing suggestions for debugging, generating documentation, automating testing, and optimizing the code. They can greatly assist the programmers. These prompts can also be useful for tasks such as code summarization, code similarity, analysis, and code completion. This enhances the productivity of the programmers and the accuracy of the programs. Here's a high level overview of using prompts for programming. For code generation, GPT based large language models can generate code snippets and scripts for various programming languages. These models produce high quality codes. These are synthetically correct and follows the best practices because it has learned from multiple sources. Debugging. Advance prompt can be used for debugging a code by providing suggestions for fixes or identifying potential errors. These prompt save time and increases efficiency by helping programmers quickly identify and resolve the box. Automated testing. GPT based language models can also be used generating test cases, test data for automatic testing. This helps programmers identify potential bugs and ensure the quality of the code. Chain of thought prompting is a powerful tool for advanced prompt engineering and Chat JPT. It is a technique that involves generating a series of related prompts to guide a conversation or a brainstorming session. These prompts encourages the model to continue generating text in a specific direction. This results in a more coherent and contextually relevant responses. Let's look at three examples of chain of thoughts prompting. Word association prompts. These involves generating a list of words relating to a central topic or a theme. Using these words as prompts for further exploration. Example, if the central topic is food, the list of related words might include cooking, eating, nutrition, or restaurants. Idea expansion prompts. These involves building on the initial idea by generating a series of related prompts. For example, if the initial idea is to create a new product and the related prompts might include questions like, what problem does this solve? Who is the target audience for this product? Scenario prompts involve creating a hypothetical scenario and then generating a series of related prompts to explore that scenario in more detail. Scenario prompts involve creating a hypothetical scenario and then generating a series of related prompts to explore what scenario in more detail. Example, if the scenario is the world without electricity, the related prompts might include, how do people communicate? How do they stay warm? How do they access information, and so on. Let's now explore how chain of thought prompting can be useful in a different context. Brainstorming sessions. If you're driving lean Six Sigma projects, chain of thought prompting can be used to generate a wide range of ideas quickly. By using a series of related prompts, you can explore a topic from multiple angles and come up with new ideas that we might not have considered otherwise. Let's think about interviews. Chain of thought prompts can be useful in interviews to explore the candidate's thought process, problem solving skills. By using scenarios or idea expansion prompts, interviewers can gain insights into candidates ability to think creatively and logically. Problem solving activities. Chain of thought prompting can guide a team towards a solution by breaking down a complex problem into smaller parts. Using idea expansion prompt, team members can work collaboratively towards a shared goal. Overall, the chain of thought prompting is a versatile technique that can be applied in many different scenarios. This helps in creating or generating ideas, exploring complex topics, and guide conversation towards a specific goal. Let's create a chain of thought prompting, using Chat GPT to demonstrate an example. Hello. Welcome to this demonstration on creating a chain of thought prompting, using ChaGPT. Chain of thought prompting is a technique that involves developing a series of connected questions or statements to guide the conversation and with AI allowing for more in depth and focused discussion. In this demonstration, we will explore how to develop an effective chain of thought prompt and response with a focus on Python programming. We will assume that you have some familiarity with Python or other similar programming language. To get started, open a web browser on your computer or a mobile screen or navigate to the chat dot.com. Once you are there, sign in into your OpenAI account. This feature is also available in the free services, but the limitation is there. Chat GPT is busy during the daytime. Establish the context is the step one. To begin, we need to establish a context for the conversation. We will do this by starting with a simple prompt. That's to introduce the topic of Python programming. Type the following in the prompt text box at the bottom of the screen. Let's discuss Python programming, specifically building a web scraper. No need to respond immediately. This is just a setup of some context. Step two is to create a sequence of related questions. Now that we have established the context that we want to use Python programming for web scraping. Let's create a series of related questions to guide the conversation. What is web scraping? Why is it important in Python programming? What libraries are commonly used for web scraping in Python? How can we install these? How do I import these libraries into our Python script? Can you provide some basic examples of web scraping using Python these are the questions that I have asked. Step three, input the first question. We will now input the first question into chat GPT and allow the AI the time to generate a response. Remember, to include the context in the prompt. In the prompt text box, type, let's discuss Python programming, specifically building a web scraper. What is web scraping? Why is it important in Python programming? Chat GPT response is shown on the screen. Web scraping is the process of extracting data from website. It involves making HTTP request to the web page, downloading the HTTP content, and then parsing the content. This helps in extracting the desired information. Web scraping is commonly used for a wide range of applications, such as data mining, data extraction, and data analysis. Also for web automation. Python is a very popular programming language for web scrouping due to its simplicity, readability and extensive library support. Some key reasons why Python is widely used for web scraping are easy to learn. Python syntax is like straightforward English, making it excellent choice for both beginners and experienced programmers. Extensive library support. Python has a rich ecosystem of libraries that simplify web scraping task. Popular libraries include beautiful soap, request, scrape, selenium, et cetera. Flexibility. Python is versatile enough to handle both simple and complex web scraping task. Community. Python has an attractive active community that provides support and shared resources. This completes the initial step of building a chain of thought prompt with Char GPT. We can continue developing our sequence of questions and responses, dwelling deeper into the topic based on the responses generated. This process helps guide the conversation, making it more structured and focused. 16. Prompt using comments: Now, I may have some kind of prompt using commands. But this time, I want to generate the code in Python. In order to ensure that I do get the response in Python, I use a leading word to nuch the model in the right direction. Notice the word Import at the very bottom. This is the leading word indicating that Python code is what I'm looking for. Well, this time around, when I ran it, Chat GPT did not get it right, but the last few times I did get it right. So remember, prompt engineering is not an exact science. You have a trial and error. Tweak your prompts until you get the right response that you are looking for. When you work with generative AI tools, it's not just about code generation for the scenarios that you specify, either using comments or using just prompt in plain English. It's also about leveraging generative AI tools to help you with coding in other ways. For example, let's say you're doing a migration project. Originally, the entire project was written in Python, but maybe we are migrating to the Java programming language. Now, this migration is unlikely to be straightforward. I'm sure there is a lot of nitty gritty that we have to take care of, and that might be a certain function that can just be a lift and shift. So you convert it to Java from Python. But when you use generative AI, this technique that I suggest here will definitely work for simple projects and simple functions. But I will also work with fairly complex functions to show you how it's possible. So there are things that you need to try out and see whether the things are working out for you or not. Now, I have some code in Python. I want to convert it into Java. This code takes the input arguments, the name, and the age of some students and convert this information to be a representation in the form of a dictionary. Where the name is the key and age is the value. You can see there are some error checking at the beginning of the code as well. We initialize a student dictionary and here is a for loop that I run over the names to initialize the dictionary. Now, let's see what the Java code looks like. You can see that a CHAP JBT has correctly mapped the dictionary data structure in Python to the hash map in Java, which is great. It has also set up a class within which it is placed the Java code. The class is called as a student dictionary creator. Again, a very meaningful name is given. Then there is a main function where you actually invoke create student dictionary function. If you scroll down below, there is a function itself. You can see that there are some error checkings like if the name dot length is not equal to age dot n, that's great. The code then initializes the hash Map, which is the right data structure to use corresponding to dictionary in Python. Now, the basic setup for the name and the age of the student is done. So this is a great conversion from Python programming language to Java programming language. Now you can convert structures that are available only in Python to Java code as well. For example, list comprehension is something that is available only in Python. There is no Java equivalent for this, but you can see that our list comprehension code has been correctly converted to Java. The original code basically filters elements from the original list, so we can only have elements greater than 100 in the filter number list. And you can see that Chat GPT has correctly telling you, here's the equivalent Java code for the list comprehension. And it basically writes a very long detailed code to give you the filtered numbers in Java. Now, the code conversion works from and to many languages. Here I will show you a few more examples. But really, hat GPT understands many programming language, and you can convert the code between all of these. Here, I convert the same student dictionary code to JavaScript. You can see that chat GPT will give me accurate results in JavaScript as well. Now, one thing about getting Chat GPT to convert code and even generate code. Make sure you double check your code and you perform your own test. It's just like you would do when you take code from stack overflow. From another site that helps you debugging the problem. You will always double check the code. Remember the concept of human in the loop. You should do the same with chat GPT. Here, I convert the same code from Python to C plus plus. Once again, chat GPT is successful. It's pretty clear that there are a variety of programming language from which you can use chat GPT for coding. And converting that code and generating the new code in the other programming language. I won't walk you through all these generated code, but you can see that Chat GPT has correctly used the vector to represent the input arguments and pass to it and an unordered underscore map as an equivalent of a dictionary data structure in Python. Now, I don't know if this was happened to you before, but this often happens to me. There are some complex bit of coding I need to do and I'm not really able to wrap up my head around the problem. And then I turn to stack overflow or a site like that that can help me out. I get some code from that side. It seems to work, but I don't understand what that code is actually doing. And because I don't understand, I'm not able to tweak it, or I'm not able to modify it to my usage. And this is a problem I am facing many times. Did you face any problem like that, please mention in the comment section. Coming back. Well, now I have my generative AI to help me out. You also can use it. This generative AI tool can explain a bit by bit code to you as well. I have some code here which I have taken for extracting the list of numbers, picking a number at random from that list, and based on the number selected, it performs one of the three operations, return zero, returns square of the number, returns the cube of that number. Now, I'm going to ask Chat GPT, what is this code doing? You will see that my instruction prompt here is, please explain me process random number does. I want to understand in simple English, and it produces a very incredible results. Now, this particular function was fairly easy one, but Chat JPT got it perfectly right. You can use this explanation to wrap your head around what the code does and then tweak it based on your use case. Let's try this once again. Here's an example of the random function. This isn't Java. I am personal I have purposefully not given this function a meaningful name so that I don't give away what the function does to ha GPT, using the name of the function. And then there is some code here that uses the string builder. This function reverses the words in the sentence that you parson. Now, I'm not going to use the term reverse anywhere in the function. I have asked Cha JPT. Please explain what do the random function does. Let's see what hat GPT is able to tell us about this function. You will see here that Chat GPT gives us line by line explanation of what the code does. But it also figured out that the code is trying to build a reverse line. The term reverse wasn't mentioned anywhere in my input. The fact that I am really careful about that, but still the chat GPT figured it out. It is now giving me explanation about the code. Maybe it is enough for me to get started. Another thing that happens with me really often is that when I'm writing the code is I have written a complex bit of code. There is one small thing that went wrong, but I just can't figure out what that is. I'm trying different inputs and keep debugging my code. But basically, my brain freezes and I'm not able to debug it with myself. Well, now I have my generative AI assistant to help me out. Chat GPT, bad or any other AI tool can help you debugging your code as well. Again, I would be very careful with any Coda debugging that they do. I would always double check it, but it's very, very helpful because they can quickly get to the point where the problem is. They quickly identify the errors that is irritating me and end up wasting my time, maybe several hours. Now, Chat GPT or the GEI tools are available at my rescue. Languages that are powered by Char GPT or GPT 3.5 on the open AI playground, I'm going to task this model to help me debug this bit of code. I have here a code that is in infinite loop. I didn't indicate the language the code is. I just pasted the code and I said, can you please interpret this why does this code not complete? Well, it doesn't complete because there is an infinite loop is the answer that I received. Let's see what Chat GPT 3.5 has to say. You can see that this model has been so good about everything I have seen so far, essentially a brain freeze and gives me something which doesn't make any sense. This code does not complete because it's missing the closing parenthesis, the colon in the V et cetera. Well, I think it has also now got confused. So let me send a clarifying prompt. This code is in Python, and now it comes back and apologizes, gives me the answer, which is, again, not perfectly right. Let's separate the answer into two parts, the text and the code. You can see the chat GPT is always very apologetic when it gets something wrong. And the GPT 3.5 model, which powers chat GPT is also apologetic. But it does not mean that getting things right just because of the apology. It now talks about missing an indentation before the print statement, which is wrong. I did have an indentation, but it gives me the correct code. It adds count plus equal to one. So I no longer have an infinite loop. I got the code part right. I got the explanation wrong. So it did help me, but it didn't get everything perfectly right. Let's see if Bard can solve this problem. I am going to switch over to bad. Ask the same exact question. Why is this code not complete? The Bard gave me the great answer. The code you provided does not complete because the il loop is infinite loop. Notice that this is the very first line. It then gives me the explanation of where to go and fix this. It actually gives me the Python code that can fix this. This answer I found to be more useful. For more examples like this, continue my lesson. Thank you. 17. Hallucination in AI Models: Hello, everyone. Welcome back. Today we are diving into fascinating and slightly funny topic, hallucination. And no, I'm not talking about hallucination in humans. I'm also talking about hallucination in AI. Or I should say, I'm just going to talk about hallucination in AI. You might be wondering, wait, can AI hallucinate like humans? The answer is yes, but in a very different way, by the end of this class, you will understand what hallucinations are and why they happen in AI models. How we can tackle them, and let's jump in with an example from the K drama. I have been watching. So when I was watching this amazing Korean drama on Netflix, in it, there was a girl who keeps seeing whales flying in the air. These whales don't exist in real life. They are part of her imagination. The term that immediately come to my mind here was hallucination. Now, hallucination happens in humans when we mix reality with illusion. For instance, someone with bipolar disorder or someone who's sleepy and under influence of certain drugs might hear and see things that might not really exist. It's like being in a dream when still awake. We are mixing the real world with the imaginary thoughts. Interestingly, AI models like hat GPT, ama also exhibit hallucination. Trust me, this is one of the biggest challenge in generative AI today. Let me explain with a couple of examples. Alright, let's start with the first example. I came across a video on Wikipedia page for hallucination. It was generated by open air video model Sara. The video depicted the famous Glen Finn in Scotland. You know, the one from the Harry Potter movies where the GodbordEpress crosses a bridge. In the real life, adtt only has one truck, one track. But in the generated video, the AI created the second flag out of nowhere. And to make things even more bizarre, it added an extra chimney to the train. The real train has only one chimney, but the AI hallucinated and gave it two. The another example closer to is when I asked an AI model, who owns this poaching center? At first, it correctly said it belonged to me. But then I pushed back and said, No, I think someone else. The AI immediately apologized and confidently replied, You're right. It is owned by Bindin. Now here's the funny part. It is not only invented a completely wrong answer, but it went on to add Bin Ladin in a dead as a dedicated educator and a software developer. This is completely made up. The EI confidently presented false information in such an elegant and believable way that it did not know the truth and you totally fall for it. So why does this happen? Why does AI model hallucinate? Let me break it down into three main reasons. Predicting pattern versus true understanding. Imagine a chef who has never cooked before, but has read hundreds of recipe books. This chef knows how to write recipe for any dish but has no real cooking experience. AI is semilar. It doesn't truly understand what it's saying. It's more like a schastic parrot, meaning it predicts pattern based on what it has seen during the training period. For example, I asked an AI model to create a recipe for a wheatgrass key. Now, the wheatgrass key doesn't exist. It's not a real dish, yet AI confidently made up a recipe describing it's a healthy quest of the traditional Indian dessert. The problem here is that AI isn't drawing from the real world experience or a true understand. It's just combining patterns it has learned often leading to nons. The second reason is insufficient data or lack of fine tuning. Hallucination also happens when the AI lacks proper training data. Let's go back to the weed grass here example. If the training data had included information like weed grass is only used for juices and not for desserts, the AI might not have made that mistake. Contrast that with tobacco, if you ask AI for a recipe using tobacco, it will respond. Sorry, tobacco isn't safe for consumption. Why? Because the training data included information that tobaccos harmful effects. So the AI knows not to suggest using it in the recipes. The key takeaway here is the quality and the completeness of the training data matters a lot. The third factor for hallucination is ambiguous or incomplete proms. Sometimes hallucination happens because of the way we ask questions. For example, I asked the AI model. Tell me about Bloomberg. Now, Bloomberg would refer to the company or Michael Bloomberg, the person. When I followed up with what is the A, AI got confused. Instead of answering with Michael Bloomberg's A, it says it was released to the public in 2023. Clearly the AI misunderstood my ambiguous question and gave an incorrect response. If the prompt we provided are vague and unclear, the AI is likely to hallucimate. Now that we know why AI models hallucimate, let's talk about how we can tackle this problem. Representative dataset, AI model needs huge volumes of high quality data covering a wide range of scenarios. For instance, if the training data includes detailed information about wheat grass, eve and their proper uses, AI will make fewer mistakes. I cannot say never make a mistake. I will correct myself and say, AI will make fewer mistakes. Fine tuning and validation. Have you noticed the thumbs up and thumbs down button tune that is available along with Chat JBT? This feedback mechanism helps the developer continuously improve the model by correcting the errors and fine tuning it. Over time, it reduces the chances of hallucination. Knowledge based system. One effective solution is to combine AI with knowledge based solutions like database for frequently asked questions. For example, if you are building a customer service chatbod, you can restrict it to use only the company's database for answers. This prevents the AI for making up things based on general knowledge. Prompt engineering. Finally writing clear and precise prompts can go a long way in avoiding hallucination. Instead of asking vague questions, be specific. For example, instead of saying, tell me about Bloomberg, ask, tell me about Michael Bloomberg, the person. To sum it up, hallucination in AI happens when it confidently generates false information. The main causes for the pattern production is insufficient data and ambiguous props. You can tackle this issue with better dataset, fine tuning, knowledge based systems and clear proms. Let me know your thoughts and keep yourself as curious in learning. M 18. Prompt Engineering for Gen AI: Think of chargebty and similar tools as the way to shape and explore your ideas. They may help you come up with and refine creative solutions. Not just about writing essays or answering questions. Here's an example. I have an idea to create a meal plan combining different cuisines while keeping it keto friendly to ease it by making the ingredients with the regular grocery stores. I asked Chad Jp to help me with this. It quickly gave me a meal plan with breakfast, lunch, and dinner ideas that would fit my criteria. For breakfast, it suggested scrumbled eggs with onions, tomatoes, and some spices. With a note that how to replicate the spices using common ingredients. For lunch, it recommended some Indian style kababs with grilled vegetables and Cucumber tomato salad. I then wanted to make a meal plan more appealing for my 9-year-old daughter who might be picky about how the new food is. I asked Cha jab to create some Pokemon stories related to each dish to make the meal time more fun. These stories included cliff hangers to excite my daughter about trying the new food. Chagpt can help turn it into a math game like involving Pokemon to teach my daughter how division and nutrition can work together. It asked questions, provided hints, and gave feedback on the answers. Finally, I thought of tuning into a new web app and Cha PT helped generate the code for that too. With Chad GPT, you can start with basic ideas, quickly help them into detailed meat plans, engaging stories, educating games, and even softwares. This example shows how versatile and helpful this tool can become. Now, let's understand how this with more examples that is showing the breadth of their capabilities. Let's start with a fun and a practical example where I wanted to create a meal plan which should ensure that it has a daily intake of 2000 calories. I asked Chad GP to help me design the meal plan. Within moments, Chad GP provided me with a detailed plan for breakfast, some scrambled eggs with sautee onions, tomatoes, and some herb spices. Which includes some chili powder, paprika, garlic powder, ginger, cumin, and coriander. For lunch, it advised me to go with bobs with grilled vegetables, bell pepper, zucchini, and egg plant, and give a side with a tomato and cucumber salad, dressed with olive oil and lemon juice. For dinner, it suggested a nice Calfloer rice with d, which mimic the texture of a traditional rice with low cubs. GPT does not have to give the meal idea. It also provides approximately serving size to ensure each meal fits within the 2000 calorie limit. For breakfast, it suggested that I go with two portions and a half cup of onion and a half cup of tomatoes and one teaspoon of the herbs that could serve a medium avocado. If you want to make the meal more enjoyable with the family, especially with pica chill children at home, you will take help with ChargeP. For example, snack idea, keto friendly snack ideas, combine flavors from both cuisines like spice, not, vegetable sticks with yogurt baste tip. Educational games for kids. Suppose you want to incorporate educational games ideas during the meal time? Given that my daughter loves mats, I might ask a gift to relate a math game involving Pokemon and nutrition. The math problem, Paco and friends were preparing meal for their Pokemon party. They had three cups of berries and need to divide them equally among four Pokemons. How many berries will each Pokemon get? If they want to add one more cup of berries, how many berries will each Pokemon get now? The hint and explanation given by Cha GPT was helping my daughter understand the math involved. For example, to divide a three cup berry into among four pokemons, you will need to divide three by four. Each pokemon will get three quarter of the cup of the berry. Let's learn one more example of crafting stories for family engagement. Let's consider another scenario where I wanted to create an engaging bedtime stories for my kids, and I use Cha jet to generate personalized stories based on the interest, adventure stories. Your children's favorite superhero and their friends are a thrilling adventure. Each chapter of the story presents a new challenge that ties into the favorite. Now, let's do a practical example of doing these stories together. I'm going to share the prompt. Yes. I'm on the Chat GPT screen, and here is my first prompt. Where I said, The first example was about creating a fusion meal. The problem that I'm using is design a meal plan that combines two different cuisine, Italian and Japanese, while keeping it keto friendly and using ingredients that are easy to find typically in a grocery store. In India. Remember, I am vegetarian and prefer no eggs and meat in mine. The meal part should include breakwat lunch and dinner ideas, and this shows what will happen. As soon as I give this idea, it was happy to create a detailed breakwat plan for me, keeping in mind my ingredients. It has given me exactly how many calories each of the item will have, and what would be the serving size and what will the total calories. For the lunch, it suggested what are the ingredients and what is the serving size and the total calories for the dinner and the total break up for the calories for the day. They have left some weeks of snacks in between so that I cannot remain hungry throughout the day. I can just say give me the making of recipe sigue. Now, it says, to prepare Zukine noodles, use this and peel of the Zukine into noodles for making the dressing to assemble and so on. It has given me a detail detailed instruction, how I can make the miso soup and so on. Now, let's go to the second example of how I can create crafting stories for family engagement. Here, my prompt, as you can see, is about developing stories. I click on Enter and it starts off with an introduction, the adventure, the hero journey, and the resolution. The time traveling twins. It gives me a summary, and so on. This can ensure that I can keep telling short stories to my kid whenever I want to keep them engaged. Now, let's assume that I want to plan for a themed event. To plan for a themed event, I need to go ahead and give a prompt like this. Plan a themed event like example, a pirate party or a science fa, and outline all the details needed for the successful event. They should include ideas for decoration, games, and menu themed food. Remember, I am vegetarian and based out of India. It says space acy when it did the background, the hanging decoration, the table settings, entrance, lighting, and so on. The game activities it said very clearly, marketing the draft stations, Alien hunt, space Bilo, astronaut training, what material is required? What are the instruction? For example, I like the space training obstacle course very well. Give me a detailed instruction for this and what are the dos and dos for the gain. Keep in mind, we have kids from nine years to 14 years. I am the only parent at the party. I've become more constrained by giving them instruction that what should be the dos, what is the age group of the kids, and how many adults are there at the party. Then it tells me, what are the material needed instruction and safety checks. This is very important. As it helps me to put up what to do, what to avoid. Avoid overcrowding, avoid h, don't allow rough people. Don't ignore injuries, don't scape the safety cares. I will be pasting all these prompts for you in a very simple language in the attachment. Suppose I want to personalize a work project. I can draft a project proposal for an initiative like a marketing campaign, and then add more constraint to it. Include objectives, strategies, budgets and expected outcomes. Provide a detail plan that outlines how to achieve the goals and measure success. So the proposal is clear and simple English is used. It be motivating and persuasive for the client to be. Don't worry about the spellings. It is going to fix it. So I have the introduction. I have the objectives to drive sales growth, enhance the strategies, digital agency, e mail marketing, event marketing, promotional offers. What would be the bet expected outcomes. Now, the budget it is in dollars and not in INR. Redo the budget in INR. I am doing this event in Mombai. It has now broken it down into Indian rupees, giving an approximation and then keeping in mind that this activity is going to happen in Mumbai. It is also giving me that, I can do content creation, influencer partnership, social media, PPC campaign, retargeting ads, launch the events, promotions, and so on. This is giving me a very interactive way of learning and not using Chat GPD only as Google. Thank you. I'll see you in the next class. 19. ChatGPT Starter Code Prompt stackoverflow: Another thing I have done while coding is if something does not work or I have broken my head trying to debug it, I still am not able to figure out what's going wrong. I turn to a colleague. So if someone isn't available, I turn to stack overflow. And I go in and paste my code there and request somebody to respond for debugging issues. And I simply copied the question over, and now I'm pasting this question into bad I have no idea what this code is all about, what the person is writing in the code is trying to use this code for. I just got the question. I saw the error in that question. I saw that it has been answered and the error has been fixed in a certain way. And I simply copy pasted the entire question. I took the code from the question, and here is what I asked Bad. Running this gives me an error. It is an unsupported operation exception. Can you tell me why? And can you fix this code for me? I should tell you that the original question had a lot more detail. Here, the exact line number where the error occurred. There was present in the stat trace. I just got rid of all of that and I wanted to see how bad does this with this very plain code and a very simple question. And I must say, I was pleasantly surprised on how accurately Bard pointed out the issue. There is an array dot as list this call somewhere in the code, and that essentially creates an unmodified array. The programmer was then trying to modify that array, which was resulting in an unsupported operational exception. And the stack overflow supporters and responders got it right as well. But so did bad. But not only explain me to fix it, very intelligently, it gave me the complete correct code. With the fixed included. Very, very useful if you are debugging issues. Now let's try going to stack overflow, and I identify another debugging issue. I will show you how I just copy pasted the question into bad and got the response. So it's a new B question, but not printing correct values. What I'm going to do is simply copy the entire question over so that I am not worried about actually understanding the question. I will look at the response, and then I will see whether it makes sense or not. Let me just copy the entire question over. I have not changed anything here. It's just the question as it is from stack overflow. Now I'm going to paste this into bad. And hope that Bard can help me. Now, this last paragraph here, I'm going to get rid of that. This is essentially a question asking for suggestions for programming. I just don't want Bard to give me suggestions. I want Bard to debug my code. Hence, I'm removing that paragraph. The new B has set up some bank account class with withdraw and deposit methods. And here, when he calls withdraw, the sum ends up getting added to his bank account instead of being subtracted from his bank account. And that is his problem. He pasted a lot of codes to help debugging and find the solution, and Bard got this right right away. The problem is that the withdraw method, it has to do is to double minus sign. The freebie has specified that it should be the balance equal to minus. If you remember the basic mathematics, this ends up being a plus operation. That's why the amount that the newbie was trying to withdraw was getting added back to his balance. In addition to pointing out the mistake, Bad also gave me the correct code so that this is very useful. But of course, at the bottom, you have the complete code as well. Let's see how this compares with the response of the Newbie got on the stack overflow. I'm going to switch back to stack overflow where I got the original question and let him give the response and show you the response that was the right one. You can see that this response was double negative as well as gives the fixed code. Essentially, when the Newbie got from the stack overflow, the newbie could get it using a generative AI tool. Let's do this once again. I'm going to go back to stack overflow, and I have found another different question. Again, a Newbie question because these are easier for us to understand and here the purpose of this demo is to help you understand how to reach out to Chat GPTR JN AI platform. I'm going to essentially copy this question over and let's see if Bard can solve this problem. I'm going to just copy the question. Now the problem is this that the Nubi gets some kind of type error while running the Python god. The Newbie knows which line causes the error? It says that the line with sum in brackets W, but doesn't understand why that error exists. Now to fix this, we have just asked Bard, see if Bard can fix this for us. And indeed, it did it right. The variable W was obtained by an iteration over the four loop. The variable W is an integer. The integer objects are not itertable. You can see that the Python code that the bad suggest is actually getting the right results. Now I can tell you upfront that it is indeed correct. What I did was to look at the code. When I looked at the stack overflow and saw that the right response was already given, BAD also got it fixed perfectly. You can see that the write response on the stack overflow and the response suggested by the Bard is exactly the same. So Chat GPT, bad or any other generative AI tool can help you debug more complex problems. And now, I suggest you to try these out on your own, pick up some of your own coding. But my response to you is they won't always get things right, so it's better to not rely on them blindly. Thank you so much for watching this video, and we will continue in the next lesson. 20. ChatGPT Starter Code Prompt Zerodha API: C. Let's continue understanding how we can use Chat GPT and generative AI to solve our coding problem. In the last few parts of my demo, we saw that we write a code using generative AI tools. Let me show you how hat GPT BD or any other generative AI tool can help you with documentation and then write code based on the documentation. In India, there is a very popular trading platform called ZeroDA. Zero DA offers an extensive API that can use for interacting with the stock market here in India. Here I ask HAGPT a question to see whether CHARGPT knows anything about zero the API. And HAGPT very clearly tells me, I'm sorry, but I can't assist you with that request. So CHAT GPT has no idea what I'm talking about. Now let me go ahead to zero the API documentation. So I'm here at kids dot trade slash DOGS Connect VR Orders. Keeping in mind CHAT GPTsRquest length limitation. I'm going to copy over this entire documentation until a certain section. So everything up to that, but it's not including retrieve orders. Once I have copied this over, I'm going to paste it into hat GPT. I'm going to ask a question based on this documentation. Ready? Here's my question. Using the documentation on this page, can you write a request in Python to place an order on Zero Da? Here's the exact same question as before. But now I have the documentation that Chat GPT can refer to and figure out how to use a Python code to do this activity. Now let's go ahead and submit this. We will see that hat GPT has to say. Chat GPT server were overloaded. So now I'm recording this again. It gave me this error and asked me to regenerate the response. When I regenerated the response, I got something that was also correct. Now, I'm not sure whether it is 100% correct, but I'm actually trying to see whether it works or not. I actually pass the documentation and show you a way. You can place an order using the request API. Now, BAD also have access to live documentation. Remember, BAT supports scraping the URL on the web to get the content and answer the questions based on that content. So I asked Bard to write a request in Python to modify an existing order on Zero DA. And Bard knew exactly what I was talking about. And it gave me code. I looked at original. I looked at the original documentation. This code seems absolutely correct. So it doesn't matter whether you're working with HRGPT, bad, Bnchat or anything. They have the ability to see the documentation and give you the code that you are looking for. So I would request you to pick up some of the documentation that are available with different APIs like the Google Map API, the weather API, and try doing the coding. Thank you, and I will see you in the next lesson. 21. ChatGPT Generated knowledge prompting Augmented: In this demo, you will work with what's called a generated knowledge prompting or augmented knowledge prompting. It's important for us to learn the new technique. Now the term actually describes what is this prompt about? You augment the knowledge of the model with some additional information that is relevant for your business or your use case. Then the model will try and use that information to respond to your prompts. This generated prompt or augmented knowledge prompt is the knowledge that the model did not have before. It can be the knowledge that's specific to your organization. For example, if you are building a chat board for your organization, you might want to train your model with some augmented knowledge about what's the company policy. And then this augmented knowledge will be used by the model to respond to your queries within the chatbot app that you have just created. Here's a simple example of how you can set up augment and knowledge prompting. Now I have this entire scenario drawn up here. Consider that you are an assistant to a computer science tutoring center. Now I want the model to act like a chat booard in a computer science tutoring center. The objective of the chat booard is to help students who log into their queries to find the right teacher to help them out. Students will message into the chat booard and describe the problem that they have. The chat booard will try and assign the right tutor to actually help the student. I have a series of steps that the chat booard should follow in order to direct the students. Let's look at each of these steps in turn. Step one is, ask the student what the problem is. The student might log in with hi or hello, and the chat booard should ask the student and try and gauge what the problem is so that the right teacher can be assigned. The chat board should then specify and identify the topic of the problem. See whether the problem belongs to any of these categories. Here are all the subjects that were taught in that tutoring class. Java programming, Python programming, C plus plus, SQL, machine learning, GCP, AWS, Azure, data analytics, data science, and so on. On the right category of the problem has been assigned, then we can go to the Step three for the chat booard. The chatboard should then try and identify the topic area and assign the right teacher to help and solve the problem. Now, different teachers deal with different topic areas. You can see the list here. For Java programming, there are three teachers, Alice, Mary, and Bob. For SQL, we have two teachers, John and Jane. For machine learning, we have Rome, Shuda and Yang. For AWS, we have Jasmine and Raj. Now, remember these teachers can only help with their problem areas. So you can't assign a student who has a Python problem to John. Once the right teacher has been assigned, then go to step four. If the student ask for teachers explicitly, check to see whether the teacher teaches that particular area. If not, let the student know. And if the student asks for their favorite teacher, but they are not the teacher for that particular subject or the area, then the student should be informed. If the teacher does indeed teach the subject, then we will assign the student to that teacher. Finally, step five, the chat board should wish the student luck for their learning journey. Now, this is a prompt that I'm going to copy over. It's a huge prompt because there is a lot of information I wish to convey to the chat board. Now, let's switch over to chat GPT and paste this prompt in. We have already seen all the details of this prompt. Now it's for us to see whether hatGPT is able to follow all of these augmented knowledge and the direction that we have given here in this prompt. Chat BGPT gets back to me with a query. I can help you. Please provide me with a problem or the topic you would like assistance with. Remember this was the step one. I have a problem with my code. Could you help me? Let's see how my chat booard assistant gets back with it. I would be happy to help. Can you specify the programming language or the topic? This will help me direct you to the right teacher. And then maybe the student comes back with something like this. I'm trying to code in Python. Let's see if the model understands this. Great. If you're looking for help with Python programming, you will see that it's now trying to identify the Python teacher, and it comes with two options. Christine and Richard. And yes, both of them are Python teachers. Are they? You can double check your prom to see whether it's indeed true. I did, too. Christine and Richard do teach Python because I mentioned it in one of my prompts. But maybe the student has a favorite teacher. I would like to talk to Kisha, but Kisha doesn't teach Python. Let's see if the chat booard gets it and what does it do? The chat booard comes back and says, I apologize for the inconvenience, but Kisha specificly helps in GCP and not Python programming. And then the bot requests them to choose one of the Python programming teachers because they would be the right one to help. Now maybe the student responds something like, got it. Please assign me to Christine, and then the chat booards work is done. The chat board correctly wishes well my learning journey and then directs me to Christine. This was fun. Let's try this once again. I'm trying to upload a fille to the Cloud, but I'm having some trouble. Could you please help me? You can see that the chatbot identifies that I'm having trouble with cloud programming. And it tries to pick up the right programming, and it narrows down to GCP and AWS and Azure Cloud platforms. Then ask me, which Cloud platform am I using? Isn't this quite cool? Now let's say I'm really clueless. The student says, I'm not sure. But can you tell me what Cloud programming should I be using? The chatbot has a lot of patience with me. I'm glad. Let's try to figure it out. What service or platform are you using? What is the URL looking like? Which website is it? Are you trying to get this information for me? Now, you can see the chatbot is actually talking like a human. Let's get back with something that isn't the Cloud platform. The service says S three. Now let's see what the chat bot is able to identify. Oh, S three is on AWS. It's correctly figured out that S three is simple storage service of Amazon. And then it specifies that I can talk with Jasmine or Raj. Maybe the student had a favorite teacher as well. Can't Romish help me No dear, Rome specializes in machine learning, and AWS would be the best taught by Jasmine or Raj. So I would say, Oh, sure, please assign me to any of them, and the robot picks up one of them and assigns me to Jasmine and wishes me good luck. Observe that I have been very specific for the topic area explicitly, and in many cases, Chat GPT is able to identify the topic area that I'm talking about. Let's try once again. My prompt says I'm having some trouble with random forest algorithm. Can you help me with this? Now, Random Forest is a machine learning algorithm, and Chat GPT identifies this. You can see that it identifies the topic area and it finds that machine learning, and then it finds the teacher who can help with machine learning. This could be either Rome, Shuda or Young. Well, the student understands and has no particular preference. So let me respond and say, please assign me to any of and you can see that the bod assigns me to Romish. The model has no idea about this computer science tutoring center or any of these teachers. But by using the augmenting of knowledge model, we are able to set up a nice chat board to integrate with our any app. 22. ChatGPT Chain of Thought Prompting: And here we are on Chat GPT once again, ready to dive into different prompting techniques. I'm going to close the sidebar so that we have more room to view the prompts and the corresponding results. In this demo, we will study how chain of thought prompting works. At its core, chain of thought prompting is about guiding the large language model to think in a step by step fashion. This is achieved by providing the model with a few short examples that outlines the reasoning process. The model is then expected to follow a similar chain of thoughts when answering the prompt. I have mentioned before that the earlier generation of large language models were not very good at mathematical reasoning. Instead of just asking the mathematical problem and asking the model to reason it through itself, it would have actually provided a wrong answer. If we now provide a chain of thoughts or a chain of steps to the model to follow, the model more often performs better in this situation. But I have noticed that models that we are working with today automatically follow the chain of thoughts and give you step by step process for any action they perform. Now, I will first show you a prompt where I do not use chain of thoughts. And then I will switch over to the one where I will use the chain of thought prompting technique. Here, I just say the simple prompt with the customer and orders data. Can you give me the average order price for each customer who has more than three orders? I have just kind of written out what I want in English. With not much to go on, Chat GPT generates an SQUL query to perform the task that I just prompted for. The prompt was vague, and of course, the result was also not very clear. I did not really want an QuLquery. I did not have all the intermediary steps that I have been looking for. Well, it was completely my fault. I should have used chain of thoughts prompting. Chain of thoughts prompting is a technique which is especially useful if you want to ensure that the individual steps in the process uses a response that model gives are like the way you were thinking and what you exactly wanted. Here's an example of chain of thoughts prompting for the same question I asked before. So I have the instruction at the top here saying, using Python and Pandas, can you do following with step by step instructions and code? Install the Pandas library, read the two CSV file called as customers and orders. Combine the two files to form a resulting data frame which matches the order of each customer. Now, find the customer who have more than three orders and compute the average order price for these customers. Using the chain of thought prompting, I ensure that the response I get is very step by step. You can see the response here for every step I have added to the prompt, I get one step in the response with the code as well as explanation. This makes it very easy for me to use this code directly because I have been very explicit about what I want, and that means that I got exactly what I wanted here in the result. If you have a complex math or logical reasoning problem, and you feel that Chat GPT is not able to help you with the answer or any generative AI tool is not able to help you with the answer, well, use chain of thoughts prompting technique. As you saw, we give the problem or we break down the problem into individual steps and have the model give you the response using these individual steps. Now, let's say you are starting a new project, you are figuring out whether you should be using Java or Python. You could craft a simple open ended prompt like this one. Which is better Java or Python. Of course, Chagpt will give you a detailed response. The detailed response is pretty step by step. It will give you the different points that you have to consider. And then it will kind of summarize these points at the very end. There's absolutely nothing wrong with the response that we have here. And in fact, in most of the cases, it is perfect. If you're stuck on some specific aspects of choosing between Java and Python, you have some specific use cases and you want to ensure that the chat GPT has consideration of all these specific nuances, then you should have mentioned it in the chain of thought prompting. The chain of thought prompting can really help you. Notice, I have changed my prompt here. Compare Java and Python, keeping the following points in mind, and tell me which one would be better for my team to use. Step one, consider the learning curve as my team is mainly made up of junior programmers. Account for the fact that we use SQL. We need high performance and easy of maintenance. Let's scroll down, and I have a few more steps. Generate strengths and weakness for each language. Make the final decision based on these criterias that I have given above. Now, let's see what the model has to say. I won't read through the entire response that I have received. I will leave that for you. You can pause this video and read it. But I notice that each step that we specified in the prompt, it is addressed separately. This allows you to actually look at the response and use it directly because these are all the steps that you specified as part of your prompt. Then at the very end, it gives you its final decision. It supports a hybrid approach. Well, that's definitely a bit of cop up. But ultimately, the decision is up to you. Something is as important as this. I have mentioned earlier that some of the other models, especially the legacy completion model, often not perform well at math. But chain of thought prompting can actually improve their math performance. Let me show you an example of exactly this. If you look at the right hand side, you can see the model. I'm currently on the Open AI playground in the legacy completion API. The model that I'm using is text Dawncy 003 model. Now, I have my prompt here, which is just a mathematical word problem. The cafeteria has 23 apples, and if they use 20 to make the lunch and bought six more, how many apples do they have now? If you hit Submit, you will see the model actually gets the right answer. They have nine apples. But it does not show any working. It doesn't give you any explanation for why the answer is nine. And really, it doesn't always give you the right answer either. This happens at this point in time to time. But it's very possible that it just won't give you the right answer. But with chain of thoughts prompting, you can get the intermediary steps, and you will also like to get the right answer. Let's switch to that. Observe my step by step instruction. Please start with the original number of apples. Consider only the apples that were used to make the lunch and compute how many apples are left. Consider the newly bought apples and then compute how many are left. Now you can see that I have actually told the process. It actually gives you the working of the problem. The chain of thoughts prompting allows you to verify the results, answers of nine apples in ten, indeed, correct. In fact, the chain of thought prompting can be as simple as specific as an additional line in your prong. Let's think of it as a step by step instruction. Even just adding this line to show you the working and think through the reasoning process step by step. You are more likely to get the right answer. If you're working with a newer model such as Chat GPT 3.5 for simple mathematical reasoning problem, you don't need to say, let's think step by step or use a chain of thought prompting explicitly. Here, the original prompt, and without you specifically saying anything beyond this, you can see the responses automatically come step by step. The response uses the chain of thought concept. And 23. Introduction to Digital Transformation: We'll dive deeper into understanding what exactly is digital transformation. Everybody is talking about digital transformation. The learning objective of this section is that you will be able to understand digital strategies, the first step of digital transformation, enablers of digital transformation, digitally redefining the business models and the digital transformation as a concept. We will also learn how digital is applied in different industries. To the steps for digital transformation is before we begin, let's understand what exactly is digital strategy. The first step of digital transformation. Digital strategy is a plan of action to achieve organizational goals. Digital strategy is a plan of action to achieve organizational goals with the adoption of digital technologies. Digital strategy increases business benefits of data assets, including technology focused initiatives. A successful digital strategy has team members across functions, strong management, and information technology members. What exactly is a robust digital strategy? A robust digital strategy defines the current, the next, and the future steps keeping in mind the larger organizational vision. Digital strategy involves understanding the current status quo based on the experimental model and focused group central repository. Defining the organizational goals based on industry directions and competitions, starting somewhere and improving with each progression to ensure constant improvement towards digital future. 24. Digital Transformation - What and Why: Let us now understand the importance of digital transformation. The learning objectives. By the end of this module, you will be able to understand digital strategy that is the first step towards digital transformation. Enablers of digital transformation, digitally defining business models, the digital transformation definition and context, and how digital is applied in different industries. So let's get started by understanding the steps of digital transformation. Before we begin, let's understand the digital strategy, the next first step towards digital transformation. A strategy is a plan of action to achieve organizational goals. Digital strategy is a plan of action to achieve organizational goals with adoption of digital technologies. Digital strategy increases business benefits of data assets, includes technology focus initiatives. A successful digital strategy has team members across functions, a strong management, and a good information technology team. What exactly is a robust digital strategy? A robust digital strategy defines the current and the next step, keeping in mind the larger organization vision. A digital strategy involves understanding the current status quo based on experimental models and focus groups and central repositories. Defining the organization vision based on industry directions and competition. Starting somewhere and improving with each progression to ensure constant improvement towards digital fusion. Digital Transformation and digital transformation is the integration of implementation of digital technologies into all areas of the business. It results in a fundamental change in business operations and the value delivered to the customer. A digital transformation offers organization an opportunity to understand and engage the modern day buyers and deliver as per their expectations. A successful digital transformation reserves as a foundation for the smart digital strategy. Digital transformation offers organization an opportunity to understand and engage in a modern day buyer and deliver as per their expectation. A successful digital transformation rest as a foundation for smart digital strategies. Let's understand what's the gap or difference between IT strategy and a digital strategy. Let's see the difference between IT strategy and a digital strategy. Conventional IT strategies focus on long term roadmaps and budget forecast, then that extends years into the future. Digital strategies tend to rely on short term, quarter to quarter roadmap that are tied to actionable items and measured organizational pools. Digital strategy is a subset of IT strategy. Both are in alignment with the business strategy. Who is responsible for digital strategy, you might think. So let us see who is responsible for it. Usually the CIO is in charge of both customer facing initiatives and backend infrastructure support system. A few companies in chief digital officer, the CDO may be appointed to revise strategies and provide leadership with relevant data that shows how digital projects are adding value to the firm. Either way, a digital strategy should be customer focused and should address ways to improve the company's social media footprint, organic search results, customer engagement, and brand recognizon. How do you build a digital strategy? An article from HBR, Harvard Business Review suggests that using the five of the following questions to ground your understanding of digital technology. Does does digital technology change the business you should be in? How should the digital tools add more value to the business you are in? Could digital technology change your target customer? Does digital technology affect the value proposition to your target customer? How do digital technologies improve the organizational capabilities that differentiate you from others in the market? Conducting a digital audit for a business. Let us understand what is the current status quo of a business. Many small scale business assume that they are already in a digital world, just because they use social media presence, but there are other effective ways to use digital to improve their business results. A digital audit is simply an inspection of all the processes, strategies, and consequences of what the firm has been doing to establish and improve its footprint in digital channel. Performing a digital audit will help you work out whether your firm is a digital learner, digitally active and digitally advanced. Benchmark your digital readiness. Let us focus on benchmarking a digital readiness. Remember, to benchmark an organization against other similar organizations in the industry to get an idea at which activities or processes are most important. Your organization may not need to be digitally advanced in all the functions, but your goal should be to match or surpass the players in the market. What are the five rules of the digital strategy? A successful digital transformation must rest on the fundamentals of a good digital strategy. A good digital strategy is about making the right commercial choice to maximize competitive gains, margins, the value add, and then implementing it with a. Let's see all the five rules step by step. Assess the strategic impact of the digital, set your digital ambitions I place big beds, build new strategic muscles, manage transformation actively. Let's now explore each rule in detail. Assess the strategic impact of the digital. An effective digital strategy starts with the right understanding of the market environment and how it is likely to change. New technologies can reshape business economics, but it is essential to think through the implications of your organization and your broader economics. But it's essential to think a broader ecosystem of customers, suppliers and partners, too. What new offerings can digital enable? What new competitors can it empower? Let's see with an example. The impact and opportunity of a digital will vary from each function and sector. Let's see this with an impact of digital in Domino's Pisa. Digital can never replace the core product pisa, but Domino's understood that digital could strengthen its operation in terms of speed and convenience. It's customer facing mobile applications, streamline the stages of placing an order and receiving a pizza with a feedback. Set your digital ambition heights. A digital strategy fails primarily due to to less rigor or lack of bigger picture. Let's see the example of Starbucks. The goal for Starbucks was that the world renowned coffee making company, Starbucks adopted digital to make up for the slow down of their store sales. The approach they took is it identified cool ways for the customer to order and pay for their coffee by developing a mobile payment app and rolling out the customer loyalty program through digital means. The result was sales through mobile app doubled. The place big bets. When should I begin or what should I prioritize? Usually focusing on two to three smart use cases results in better clarity and delivers better results. Prioritize initiatives as portfolios and implement the one with short term impact first. Quick wins will help fund the transformation by getting the necessary funds, releasing the right people needed for more strategic critical priorities. I identifying the right beds, the customer centric view that focuses on more competitive gains and creating value is critical. Let's understand with the example of John Deer. John Deer is a farming business began with an application anchored in the activities the farm laborers are trying to optimize. That is, planting seeds rightly, adding an optimal amount of nutrition and putting minimal amount of fertilizer in their crop to prevent pest and weeds. This led to a significant change and transformation. For example, the C and spray technique, which allows every weed to be recognized through a mix of sensors and ML algorithms, then they are eliminated through highly precise, remotely controlled spray. Let's see it with a video. Building new strategic muscles. A highly planned digital strategy needs new capabilities and cultural changes. A firm needs to build new strategic abilities to implement its conventional strengths to ensure that new and work together in a coordinated way. A new digital pool of resources is critical, but always of scarce. Setting up digital culture complies a successful digital strategy. It helps attract quality talent who looks for challenging engagements. Agile methodology ensures a flattened organization structure. Firms can achieve better results faster than the firm that work in the old way. The next point is managed transformation actively. IT transformation progress are linear or discontinuous. In industries with slower moving technologies, traditional top down strategy development approach works. Planning few years out in advance can take you to a failure as changes in technology and market dynamics shift more rapidly than the VUCA world. In a fast paced, unreliable sector, you will need to be more iterative approach pre planning. One might ensure balance of strategic alignment with the above with a ground level alignment from the below. A project management office which is led by digital chief digital officer can keep the pace, slowly monitoring the progress against the established goals, milestones, and metrics and signals when it's time to just adjust the course. We will see you about digital strategy in the context of digital transformation in the next lesson. 25. Applying Big Bets in DT: Big bets. Where should I begin or what should I prioritize? Usually focusing on two to three smart use cases to get better clarity and deliver best results. Prioritize initiatives as portfolios and implement the one with short term impact first. Quick wins will help you fund the transformation by getting the necessary funds and reducing the right people needed for more strategic critical priorities. And in identifying the right beds, a customer centred to that focuses on both competitive gains and creating a value is critical. John, dear. John Deer is a farming business. It began with applications and in activities that the farm laborers were trying to optimize. Life, planting seeds rightly, adding optimal amount of nutrition, and putting minimal amount of fertilizers on their crops that can prevent pest and weeds. How did they do this? This led with a significant change and transformation. For example, S and spray technique. You can find this video on YouTube. I will also upload it over here, which allows every weed to be recognized through a mix of sensor and machine learning algorithm. Then they eliminated through high resistance remotely controlled space. Building new strategic muscles. A rightly planned digital strategy needs new capabilities and cultural changes. A firm needs to build new strategic abilities to complement its conventional strengths and to ensure that new and old work together in a coordinated way. New digital pool of resources is critical, but always scare. Setting up a digital culture complements a successful distal strategy. It helps attract the quality talent who looks for challenging engagement. Agile methodology ensures platter organization structure, firms can achieve better results faster than the firms which work in the old ways. How do you manage transformation actively? IT transformation progress can be linear or discontinuous. In industries with slower moving technologies, traditional top down strategy development approach works. Planning few years out in advance can take you to a failure as changes in technology and market dynamic shift more rapidly in the BOCA world. In a self paced, unreliable sector, we will it more iterative approach to preparing plans, one which ensures balance of strategic alignments from above to the ground level insights from below. A project management office leads to a chief digital officer can keep up the pace, slowly monitoring the progress against the established goals, milestones, and metrics, and the signals when it's time to adjust the course. In the next lesson, we will learn about 26. How to submit your ChatGPT Project: As you've come to the end of this class, you will see that you need to submit a project to show that you have understood the concepts well. I have given some project ideas to you, telling that how you can do the project and submit it once for you to get a certificate. Let me give you a small demo. If I take the first question, the first question says, complete the story with the help of Chan DBT texts. And then there is a storyline that is given. What I can do is I can simply copy this. Open my dad GPD account with I have, I take up a new chat and paste it over here. So complete the story with the help of Chaddi bit. So I will just say complete the story. I will ask you to complete the story. When I do this, the judge typically will start writing and reading the story for me. Right. So I will allow it to complete. And then what am I going to do? I'm going to paste this in my project section. Let this complete and I will show you a demo how I'm going to do it. So you can see that it's very beautifully generating a good, big response because it is a storyline, right? So what am I going to do now? I'm just going to take this Snipping Tool. I'm going to copy the question. I'm going to click on the Save button. And in my Download folder, I'm going to save Project one because it's the first thing that I want to present in my project. This is my second screen. I'm going to take this again and generate and say that this is my second piece. Then again, I'm going to click on C and C project to write. And similar way, I will go ahead and complete it. Suppose if I'm doing this, I'm using GPT for completing creating a shopping list. Create shopping list list for me. Sure, I'll be happy to help you. Can you provide some more information like what items do you need? When it doesn't have the answers, it gives me a response like this. So I can take a snapshot of this and then see what. Also. I will say this as my project brief right now I have three files which I have created as part of my project file. I've opened up my skill share under. You will see it as about review, discussion and projects. So when I go to the project section, I have an option to create a project. So let me create a project. And in the project section, what am I going to do? I'm willing to give up. I'm going to give a title to my project. So I'm going to say that this is a chad DPT sample project by the teacher. I'm supposed to upload an image. So what I'll do is I will go ahead and upload the image from my download section. So I learned my project one. So let me upload that. So that is mine. What do I say? This is my cover page. It will be just a snippet and my project is pasted below. I'm going to click on the image because I have three made Files, Project 12.3. So I'm going to open one at a time and inserted to the first image is uploaded. Now let me insert the second image. Then let me insert the third image. I have completed my project. I will go ahead and submit my projects by clicking on Publish. I hope this video was helpful to you to see that how you can submit a project. If I go back down, you can see the dimples and with the teacher has submitted so a sample project by the teacher. So I've imposed record, right? So this will help you submit your project in time. And once you submit a project, you get a certificate. So just for you to get an idea, it is giving me an OOP speed, but for you you will get a certificate. Thank you. Thank you so much and I'm looking forward to your projects. 27. Thank you: And with that, we have come to the end of Chat GPT, complete course for beginners. I want to genuinely thank you. Thank you for being my students, and thank you for completing this course with me. You didn't just learn how to use ha GPT. You learned how to think with AI, how to structure the prompts, avoid hallucination, explore advanced techniques, and apply artificial intelligence in real world scenarios. That shift is important because AI is not just a tool, it's a capability, and capabilities grow with right guidance and practice. Let me share a bit about myself beyond this class. I'm Dimple Sangui an instructional designer, AI capability builder, corporate trainer, and founder of Avisa Learning Solutions. Over the years, I have worked with professionals, teams and organizations across industries to build structured learning systems across AI. Lean Six Sigma, digital transformation, and operational excellence. My work focus on one core idea, helping people move from curiosity to competency. Whether it's the beginner exploring Chat GPT for the first time, managers evaluating AI tools for their teams or organization. Building AI capability, framework, my goal is to always keep these concepts practical make it easy for implementation, not just theory. And I would love to stay connected with you. You can connect with me on Linden, where the QR code is displayed on the screen, where I regularly share practical insights on AI, productivity, continuous improvement, and digital transformation. I also share quick learning tips, tools, and structured framework through my WhatsAp channel. You can scan the QR code on the screen and join. I also lead some instructor led programs, and you will get the notice about it in the WhatsApchannel. And if you want some self paced learning programs, a visa learning app where you have additional courses, templates, certification pathway, and guided learning challenges designed to help you grow step by step. Your AI journey doesn't end here. In fact, this course is just your foundation. Keep experimenting, keep building, keep asking better questions, and most importantly, keep investing in your own capability. Thank you once again for learning with me. I will see you in the next course.