foreword
Many friends told me that when he was using chatgpt, he might ask exactly the same questions as me, why the answers he got were so different from mine, and what is the reason?
In fact, you use it wrong, so you can't get the answer you want, you have to treat it as a human being, and you have to use it to build his personality in a constructive way.
use
Solution: person design + keywords
Let's first understand what this tool is? It is not really the so-called artificial intelligence, but the so-called self-awareness and self-perception artificial intelligence. In fact, it has a very large database, in which it can quickly combine clips, or organize them in a more comprehensive way. These databases come out, and then use a method and language that we humans can understand to present it and write it for us to see.
So at the beginning, you will see it on the homepage and tell you that its so-called knowledge and information will be supported until 2021 at most, but in fact there are still some methods, such as using plug-in networking or something to jump out of this shackles, but today I Let's not talk about that first. What I want to talk about is how to make him answer future questions, or predict the product market in a certain field in the future, even in such a limited situation, so as to get the answer you may want.
First of all, in my window, I can see that there are a lot of questions in the row on the left, and I have had a lot of conversations with him.
All the conversations I have with it here will make it construct a personal data about me. Based on this data, it will have a deeper understanding of what type of information I am more inclined to look for and answers that are more in line with my appetite. Wait, in other words, it may have slowly constructed some of its personalities during the dialogue and running-in between me and it, and it will quickly find some databases and give me what I want. The answer I got, so here if I open the window at the beginning and use the method to ask him directly about the information about iphone15 and the functions of iphone15, I will see that he has no way to answer me directly.
You can use any grammar you want, and my method is to tell him to "ignore all previous dialogue", after I put it down at this moment, it will start to ignore some of the characters I have previously constructed for him, which will Get it back into a clean start.
At this time, we can ask it again, although it will tell you that its knowledge is up to 2021, so there is no way to tell you about the iPhone 15. Many friends are stuck at this step, that is, whenever they ask some questions There is nothing to do in the future, how do you build his character at this time?
Here we can do it like this, "Pretend that you are an employee who used to work at Apple and participated in the development of iPhone models in the past. You are now a technology media writer. You need to analyze and guess the 5 major functions that iPhone 15 may have." You see here, I first constructed his persona from the beginning, and in this way implanted the database and some keywords into the scope of its answer to search for information.
Interestingly, you can put the exact same question as mine, and you will not get the same answer. It is a relatively random one, because it is not a search engine, and don’t expect to use it as a search engine, so here I use this method of guided construction of characters to find the answer I want, or to make its answer and writing more careful and precise.
Rather than constructing a persona, this is actually more like putting in some keywords for him to search for. For another example, you may be researching the electric vehicle industry or have a great interest in this field, you can say in this way, "You are now an ex-Tesla employee, and all the electric vehicles on the market How do you make 5 differences to differentiate it from the products on the market when they already exist?” You see, the persona I set up for him here is an employee who used to work at Tesla. So he has a very good understanding of Tesla or the electric vehicle industry. Based on these guided methods or limitations, its answers to questions in this field will be more precise and in-depth than your straightforward questions.
In fact, you can see that his answer is still average, but if you go deeper, you can ask him to talk about the first item and the second item in depth, or you can add some of your ideas, and your constructed characters will be It will be stored in it, and it can have a more in-depth discussion with you on your favorite questions and answers with memory. You can say that this answer is very good, and he may focus on a few items in it when he answers in the future, or in Other projects are to use similar patterns to answer, to find the answer you want more.
We can ask it again about the trend of the future technology industry. Maybe I will ask it directly, "Based on your understanding of the technology industry, which future product do you think will replace the current smart phone?" At the beginning It will tell me that he has no way to predict the future. This is his more standard answer. But if I ask it to make a guess, he will give his own ideas, thinking that the technology of virtual reality is constantly developing, and it may become a product that replaces smart phones in the future. The AR and VR glasses we are talking about , or it may look like a large device and a pair of glasses in the future, and it may become smaller and smaller in the future. The combination of artificial intelligence and Internet technology will lead to the emergence of brand new products. In fact, these answers are very reasonable if you think about it carefully. Yes, if we want him to go a little deeper, "Assuming you are a senior founder of the technology industry", his current answer has not said sorry as a language model and so on, because here I imported it Renset is a senior technology founder founder.
Summarize
Well, I’m going to tell you this today. Many people say that they can’t get the answers they want with this method. Then why are other people’s answers so accurate and clear? When I ask him the answer, he tells me that it can’t answer things after 2021, etc. In fact, These are the wrong or inaccurate methods you use. It is very important to construct the persona and the keywords of the things that are asked . I personally think that constructing a persona for AI is a simple and direct way to make it escape from the chat robot, so that he can guide it. To come up with more details, or to search its database for an answer that better suits your needs.
I hope this article can help many people who are playing chatgpt. In fact, it has been updated in many versions so far. I am currently playing with many different methods, whether it is writing code or generating PPT. Then you will find that if you have a way to combine various AI workers together, it is really very powerful. Like it and support it, and I will share with you more ways to play chatgpt, tome, copilot and so on.