Information science How to create correct AI programs while learning from ChatGPT, which has contradictions and mistakes / miki Shimizu

Product description ※Please note that product information is not in full comprehensive meaning because of the machine translation.

Information Science
[Introduction to Content]
By asking ChatGPT questions about things that cannot be understood only by reading a document alone and things that will likely get angry if you ask people many times, we will learn in detail about AI's "mechanisms that cannot be understood only by the content written on the surface." However, there are many "mistakes" and "inconsistencies" in ChatGPT, so we will lead the talk about the mistakes and inconsistencies that come up and understand the things that cannot be understood only by the content written on the surface.
For example, what is "mapping to a higher-dimensional feature space"! "Non-linear features can be learned by an activation function"? "What is the convergence of the model"? "" Hesse Matrix "is what you want to do after all! If the question" What does it mean? "doesn't make much sense, we will come up with ideas and ask them." In other words, this is what you mean? "However, even if ChatGPT says" Exactly as you say! "if you don't read the explanation after that carefully, it is doubtful whether it is really good or not.
This is a book that readers can't easily reach by just reading a series of AI textbooks.
* The code is the same as that published in ChatGPT. The conceptual scheme is drawn by the author.