Abstract

How do generative AIs such as "ChatGPT," "Copilot," and "Gemini" learn, think, and generate answers? This book answers these questions by explaining how generative AI works.

However, there are no complex mathematical formulas. The explanations are simple enough for the average person to understand, so anyone with even a middle school math background can read through. It also answers simple questions about generative AI, such as "Why can they respond as intelligently as humans?" and "Why can they tell plausible lies?"

The book also provides a thorough, basic explanation of large-scale language models (LLMs) and their core technology, the "Transformer." The second half of the book provides a deeper technical understanding of the Transformer's structure and principles, as well as DeepSeek and image generation.

Author’s Information

Freelance IT writer. Born in 1970. After graduating from the University of Tsukuba, he worked at Denso developing car navigation software. After leaving the company, he worked in web production before going independent as a freelance writer. He currently writes mainly on systems, programming, AI, and Microsoft Office. His main works include "A book that will definitely help you understand the key points and tricks of Excel VBA programming" (Hidewa System), "Excel VBA for beginners", "Python for beginners" (Kodansha Bluebacks), "Python library textbook", and "How does deep learning AI learn and reason?" (Nikkei BP). His official website is http://tatehide.com/