Full transcript of “3 Years of ChatGPT knowledge in 10 minutes”. Click a timestamp to open the video at that moment. Need another video? Use the YouTube transcript generator.
0:00 I have been using Chat GPT ever since it came out. And a few days ago I posted a community post in which I asked you how are you using AI ? So the majority said that they want to use it but they don't understand how to do it. Now if you are a student or a professional and you are not using AI in your daily life then genuinely you will be left behind. I'm not trying to scare you, just telling you. So you must know how to use AI and that is what we will do in this video. First let’s start with what is the biggest mistake the general audience makes while using AI? So, its very easy answer is lazy prepping. Like let me show you with an example. You call this make me a gym plan. Now as soon as you write this, it will give you some response which will not feel right to you because it is not specific to you. Now after doing such things people say that AI is useless. No, because every product has some
0:45 parts. As if the most important part is context. After that comes the task which you want AI to do. And then come the examples. Let me show you our gym prep with a little refinement. You will understand better. In the updated version we write, I am 25 years old and I want to gain 5 kg muscle mass in four months. I go to the gym for one hour four times per week. So now you have given it this context that what is your state right now ? After this you say that here are my protein sources. So this becomes your additional infection. After telling all this, in the end we write give me a gym and diet plan to achieve my goals. This is the task, what do you want it to do. Now when you take the response of this prop, do you think it will not be better. You should compare both the responses. That first response is very generalized. While the second one is a more personalized response as you expected. Now if you
1:32 want to fine-tune it further then tell it that I follow a push-pull leg routine so that it will tell you exercises accordingly. For examples see lazy prep hoga make me a study plan. And in this also the same generic response will come. And if we write prepare with full context then we will write I am a 12th grade commerce student preparing for boards in four months. I study best in the morning and want breaks every hour. Give me a realistic weekly study plan. And from this you can see that obviously you will get a better response. That is, the basic thing is that the more details you give, the better. Now you may ask me how much details should I give? Should I also tell him that there is a lot of trouble going on in my life and that is why I am very upset. This would get into slightly out of context territory. So you should n’t do this. So how much information to give matters a lot. And for this I will
2:17 tell you four hacks related to prepping. Firstly, whatever prompt you write, at the end write “Ask me further questions to get me a personalized detailed output”. What will happen is that it will give you a response but it will also ask you questions for further context. For example, if you try this in the gym setting, you can see that it has asked additional questions and when you answer them, the output will become more personalized. Fact number two: You have to talk to AI just like you talk to your friend. If you need advice from him about buying a phone then you don't say like this, brother, I want to buy a phone. The first question the other person will ask you is: What is your budget ? What is your preference? That means you will have to tell everything only then a phone will be selected for you and the same thing happens here also. So the first thing you learned is that you have to write things in detail so that you get the output you want. Fact number three: There is a big problem in all this pomp and show
3:03 and that is your laziness. This is the condition of the generation nowadays. It will definitely take your life to write such a big post but it also has a session. You can also get the text typed by speaking using Chat GPT. All you have to do is click on voice mode here. After that, whatever you say, it will convert it into transcription and it can understand many languages. You can talk in any language you want and it also understands Hindi. And if you want to talk to it like a completely human, then that too will happen. Tell me something about me. You are a very creative content creator who focuses a lot on technology, smartphones and behind the scenes stories of the industry. You love controversial and investigative topics and always try to make your videos unique and engaging. You also pay a lot of attention to detail and quality in your work.
3:48 It is able to tell so much about me because I have fed it so much information over the years that now it knows what I am like. It's just a small cone. Many times you will not remember its outputs because while talking it becomes difficult to remember the thing. Sometimes he writes it in Hindi in the chat form but sometimes he doesn't. So you have to take care of this thing. Fact number four: If you are still facing difficulty in writing PROP then there is a solution for that too. I have also explained this in the last GPT video. You can make Chat GPT your print generator. You tell it that you are my app creator and for this it will follow a four step process. First it will ask what you want to write about. The second one will generate these three sections based on my input. Firstly, the prepare I need, secondly suggestions as to what other context should I give to it to make the prepare better and thirdly, it will ask me questions so that if
4:35 it needs any additional information then I can give it that. And we will do this until we get the best solution. If you learn to use this hack, it is a very powerful trick. Now these are four hacks. Now let me give you three experience-based tips. First of all, you have to accept that your first attempt will never give good results. You will first have to analyze the response of the question and clear the gaps and misunderstandings in it and after that you will have to revise the question that hey bhaiya I was saying it like this and not like this and you will have to do this until you are satisfied with its answer. Second tip, if you are getting some professional work done through it, then one thing matters a lot in the responses and that is the tone of the output. Like, if I have to get scripts written by him, now he knows how I write. So if I give it instructions to write it like
5:22 me, then it will write it like me. Similarly, if you want a professional tone, then at the end of the response you will write, keep the response in professional tone or you want it casual or jokingly casual, whatever. For example write a 2 minute script about phone scams. Tone casual, conversational but alarming. The third tip is that if you are preparing something like a plan, then get the output in the form of tables or steps so that it becomes easier for you to understand because it takes time to understand things in large paragraphs and we have to minimize laziness. Now you have improved your prepping skills. The second problem that everyone faces is where should we use AI? Because now everyone knows how to write emails, summarize notes etc. Things can happen beyond this as well. For example, if we look at our gym example, what can you ask of it now given all this data? You can ask him
6:07 that I have a limited budget and I have to fulfill my protein intake within that. So you describe the whole problem that this is how I workout. I go to the gym so often and I have only this much money, so tell me how can I fulfill my protein intake ? Or if I give a very fast example, I am starting a food stall. Help me name it and create a one line pitch that's desi and catchy. Another one that came to my mind was My Best Friend is a Marvel Fan and He is Turning 20. Suggest five creative birthday gift ideas under ₹1000. It may sound a bit complicated as things do not come to mind immediately because we do not use it, that is why it happens like this. But for this you do not need to think much. Its a very simple trick. Whatever you Google, take it and take it to AI. You will automatically start getting use cases. Suppose you have to plan a trip from Delhi to Manali and you need places to visit in Manali. So do
6:52 n’t Google that. You come here and ask. Firstly, your time will be saved and secondly, you will also get direct responses. So the second thing we learned is that if we use this instead of Google, we will learn to use it. Now let us look at some more AI tools because apart from Chat GPT, you can also use other AI tools according to different use cases. Like there is perplexity to the latest news and sources. Analyze Cloud works very well with large files or PDFs. If you need support for making notes or writing then Notion AI is great. And the feature of asking questions in real time by live streaming Gemini can solve a lot of your problems. And if you want to go very advanced you can also check out Hazen. And you can follow similar things in the smartphones which are available nowadays. Just one tip I'll give you. You should install perplexity in iPhone because its voice assistant series is much better. Like now, let me show you an experiment. I want to book an Uber from NDLS to Delhi
7:39 Airport. If I show you, it has given random responses. Links have been given. Basically there are links to Uber and nothing else. Now let us look at the same thing by calling it perplexity. Book an Uber from NDLS to Delhi Airport. You can see it has selected Uber and after that everything is manual. Apart from this, if you want more better props i.e. life changing ones, then you can go to Twitter and follow pages like Propt LLM and God of Propt. These have some really awesome detailed props. But along with all this you have to keep one thing in mind that AI is not perfect. Because when you start using AI so much then you should know what problems come with it. So the most primary problem is elusiveness. In simple words, it will never say no to you.
8:24 He will not tell you that he does not know the answer. He will give very confusing answers. That is why it is said that it is necessary to do some fact checking of AI's answers. So if you are watching something important and you want to avoid this thing then at the end of the form write provide sources or say I don't know if unsure. The second thing is that every chatbot has a cut off date. That is, for how long the data is available in it. Like Chat GPT and Gemini both have June 2024. After this, if you search for any event, AI will need internet access for that. Otherwise, he will not be able to answer properly. The brain of Third AI thinks exactly as per the data it has been trained to. If its data was biased at the time of training, then the AI will also give biased answers. For example, if we ask an AI to show a photo of a CEO and it might show a photo of a white male. Women or other races are not there in it. That
9:10 means the AI is assuming that CEO means white man. This can also extend to bias, gender, race, religion, or political ideology. And to fix this thing you write with the prop that give me a neutral opinion without any biases. Add all the pros and cons of the situation. And by doing this you can avoid this problem of bias. Now I made a similar video 2 years ago and its tips are still valid today. You can see it by clicking here. The rest of me is done for today. This is it from the side. Thank you so much for watching. I am Denshu and I am on leave. Till the next one. Bye bye.
