Atlas of AI

The story told most often about AI centers on algorithms, capabilities, and future potential — and almost never on the physical world those systems depend on. Kate Crawford’s Atlas of AI is a deliberate correction to that gap. Drawing on field research across the supply chains, data centers, and labor systems that underpin modern AI, Crawford maps the material costs of the technology in concrete, grounded terms: the rare earth minerals mined to build the hardware, the energy consumed by data centers, the workers who label training data under difficult conditions, and the ways AI systems encode and reinforce existing power structures rather than neutralizing them. The book is structured around a series of specific sites and case studies rather than abstract argument, giving it a journalistic specificity that makes its claims harder to dismiss. Crawford is herself an AI researcher with deep technical knowledge of the field, which means the critique comes from genuine insider understanding rather than outside skepticism — she’s not arguing against AI in principle but insisting on honest accounting of what it actually costs and who those costs fall on. At 336 pages, the book is focused and readable, but its implications are substantial. For readers who want to understand AI as a full sociotechnical system — not just a set of capabilities — this offers one of the most thorough and rigorously supported critical perspectives currently available.

Author:
Kate Crawford