51爆料

"How does AI actually learn?" This is a question I’ve been asked many times in lectures and talks. Today, the term “AI” has become deeply familiar—we see it in search engines, smartphone voice assistants, automatic translation tools, and many other aspects of our daily lives. And yet, the mechanisms behind AI remain surprisingly little understood by the general public.
 
This book, Machine Learning as Liberal Arts, was written to address such simple and honest questions. It is intended for readers who are not AI specialists—students in the humanities, working professionals, and anyone with an interest in understanding the basics of machine learning. I have tried to explain the concepts using plain language and minimal mathematics, so that even those without a technical background can follow the discussion.
 
At the heart of the book is the concept of supervised learning. This is a method in which a system is trained on a large set of data with labeled answers, so that it can make accurate predictions even on new, unseen data. Applications of supervised learning are already widespread—for instance, recognizing people and cars in camera images, detecting diseases from medical images, or converting speech into text.
 
My own research has long focused not just on the models themselves, but on the methods of learning—that is, how a system is trained. This aspect is just as important as the design of the model. In this book, I place emphasis on the structure and strategies of the learning process, without over-focusing on any particular model. I hope to share with readers the idea that there are general learning principles that apply across many types of models, and that exploring these principles can be both rigorous and rewarding.
 
The book also presents real-world examples of how supervised learning is being used in society today. These include diagnostic support in healthcare, earthquake data analysis, and conversational aids for the elderly. In each of these cases, the technology plays a role not in replacing people, but in supporting them. This perspective—that AI should serve as a tool to assist humans—is another key message of the book.
 
Finally, one of the most important themes I wanted to convey is the value of understanding the mechanisms behind the technology we use. Rather than accepting AI as something mysterious yet powerful, I believe it is important to develop an attitude of curiosity—one that seeks to grasp how it works and what its limitations are. As AI becomes ever more embedded in our society, this kind of informed mindset will become an essential part of being a responsible and literate citizen in the information age.
 
It is my hope that this book will serve as an entry point to machine learning, and as an invitation to think more deeply about how we relate to AI in our everyday lives.
 

(Written by SUGIYAMA Masashi, Professor, Graduate School of Information Science and Technology / 2025)