With its somewhat touting title, Understand everything (or almost) …its fine slice, its chain of questions, with its simplistic drawings and its little -supplied pages, let us admit that this book seemed little intended to distinguish itself from similar productions. Error. The promise of the title is more than held: with the recall of basic knowledge mixed dives into recent scientific advances, even the introduction to still open research.
Success is due to what each of the 20 questions and answers was written by different researchers (only three women), without the level and style of the book. The chapters are without frills, going to the essentials, with paragraphs reduced to one or two sentences. They conclude with a “End word” Coming summarizing the four or five pages of each subject. The numerous illustrations serve so much educational support, providing information, as aesthetic graphic elements. The whole is dense.
Read the survey: The images generated by AI and the risk of rewriting history
Read later
The first three questions sweep the story of the contemporary era, to bring us to the explosion of the generative version of artificial intelligence (AI). The following five open the hood for learning concepts, neural networks, “tokens” (language units), inference, convolution, transformators … Even the technique of the generation of images by a diffusion model is explained in a page and five drawings.
-“Federated learning”
A series of questions follow that question the “intelligence” of these tools. Are they better than us? Are they creative? What do they understand? The answers note the limits and the many issues that these new technologies raise.
Read also | Article reserved for our subscribers Why are the photos generated by AI are so much the same?
Read later
After four other chapters on applications (healthy, robotics, in research), again offering a critical look at what is done, is doable and is not yet achievable, the book ends with four more social questions. The themes of respect for privacy, security, gender, origin, social class or ethics stimulate the reader’s reflection and are an opportunity to testify to the vitality of research in these fields. Again, no simple lowering: the authors speak of“Federated learning”causality, “Optimization bias”… And, despite the reduced size of the structure, manage to slip that the term “hallucination”, very often used to designate a classic defect of AI, is wrongly.
You have 15.37% of this article to read. The rest is reserved for subscribers.
