2022.May.26

REPORTS

Report on the 8th B’AI Book Club
Kate Crawford, Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence (2021)

On February 22nd, 2022, B’AI Book Club, a book review session by the members of the B’AI Global Forum, held its eighth meeting online. In the meeting, Associate Prof. Ai Hisano introduced Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence (Yale University Press, 2021) by Kate Crawford which argues that “AI is neither artificial nor intelligent” (p. 8; emphasis in original) and reveals the power relations behind AI.

Atlas of AI properly criticizes that preceding AI research has understood AI as an  abstract “intelligence” removed from the material world and only focused on its technical aspects without discussing its social facets. To overcome these pitfalls, the author takes the approach of viewing AI in the context of broader social structures and systems, and thoroughly tracing the agents—what is being optimized for whom and who has the final decision-making authority—regarding AI. In doing so, the book succeeds in shedding light on how artificial intelligence is “made” and what social, economic, and political influences it has. In each chapter of the book, she elaborately analyzes the various different fields where resources are exploited by AI systems, such as nature and labor. Over all, the book persuasively argues that artificial intelligence is embodied and material and functions to sustain the power structure combined with infrastructure, capital, and labor by mainly referring to secondary literatures in impressively broad fields.

After Associate Prof. Hisano’s presentation, the participants discussed the book’s potential to reach a broad range of readers. As is evident from the fact that Atlas of AI has been numerously reviewed by non-academic magazines, its target audience seems to be not only researchers but also non-experts. For example, the book is written with plain and general terms and illustrates various cases of AI systems including the digitalization of affect, which makes the problems of AI more familiar to the readers. In addition to that, the tone of the book is somehow soft and neutral, and the author does not show her anger or irritation toward a particular enemy in the book. These traits appear to make the book more accessible to common readers.

On the other hand, partly because of such tendency toward neutrality and accessibility, the book does not adequately discuss some important issues, such as philosophical consideration on materiality or remedies for the present circumstances. For instance, although the author clearly analyzes the current situation where artificial intelligence depends on a broad set of social structures and reinforces existing power relations, she does not suggest a persuasive solution for the problem or an alternative strategy that we should take. While such neutral stance of the author makes the book more approachable for general audience, it might be unsatisfactory for some readers.

In spite of such deficiency, the book is a must read for both experts and non-experts since it delineates the substructure beneath AI with plain expressions. It seems to pioneer a new research field by diverting attention from the moral of AI application to the power relations behind AI and can reach new audience who has not been interested in AI research. In this sense, as repeatedly mentioned in the meeting, it also appears that Atlas of AI is worth translation into Japanese. Since Japanese AI researchers and practitioners barely discuss the social, economic, and political aspects of AI, this book can be used as a starting point for such discussions in Japan.