A Surprisingly Insightful Exploration of AI
Despite its somewhat sensational title, Understanding Everything (or Almost)…, its slim size, its question-and-answer format, its simplistic drawings, and its sparsely-texted pages, this book seemed unlikely to stand out from similar productions. We were wrong. The title’s promise is more than fulfilled: basic knowledge is combined with dives into recent scientific advances, even introducing open research questions.
The success lies in the fact that each of the 20 questions and answers was written by different researchers (only three women), without compromising the book’s level or style. The chapters are straightforward, getting to the point, with paragraphs reduced to one or two sentences. They conclude with a “final word” summarizing the four or five pages of each topic. The numerous illustrations serve both as pedagogical support, providing information, and as aesthetic graphic elements. The whole is dense.
The first three questions cover the history of the contemporary era, bringing us to the explosion of generative artificial intelligence (AI). The next five unlock the concepts of learning, neural networks, “tokens” (language units), inference, convolution, transformers… Even the technique of generating images by a diffusion model is explained in one page and five drawings.
“Federated Learning”
A series of questions follow, questioning the “intelligence” of these tools. Are they better than us? Are they creative? What do they understand? The answers highlight the limitations and numerous problems raised by these new technologies.
After four more chapters on applications (in health, robotics, research), again offering a critical look at what is being done, is feasible, and is not yet achievable, the book ends with four questions of a more social nature. The themes of respect for privacy, security, gender bias, origin, social class, and ethics stimulate the reader’s reflection and provide an opportunity to demonstrate the vitality of research in these areas. Again, there is no dumbing down: the authors speak of “federated learning”, causality, “optimization bias”… And, despite the small size of the book, they manage to slip in that the term “hallucination,” very often used to describe a classic defect of AIs, is wrongly used.
Enjoyed this post by Thibault Helle? Subscribe for more insights and updates straight from the source.
