Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence

Atlas of AI examines the material realities and hidden costs behind artificial intelligence systems. Crawford takes readers on a global journey through the physical locations and infrastructure that enable AI, from lithium mines to Amazon warehouses to facial recognition deployments. The book investigates AI through multiple lenses including labor, natural resources, data collection, classification systems, and power structures. Crawford maps out the networks of extraction - of minerals, human work, personal information, and beyond - that form the foundation of modern AI technology. Through interviews, site visits, and historical research, Crawford traces how AI systems are built, who builds them, and what they're built upon. The text moves between concrete examples and broader analysis of AI's impacts on society, economics, and the environment. The book challenges popular narratives about AI as an abstract or purely technical achievement, revealing it instead as a force that shapes - and is shaped by - existing systems of inequality and industrial exploitation. This reframing prompts essential questions about the true costs and consequences of AI development.

Readers value the book's investigation into AI's environmental and social impacts, particularly its examination of resource extraction, labor practices, and surveillance systems. Many note the detailed research and real-world examples that expose AI's material costs. Liked: - Clear connections between AI systems and environmental damage - Focus on hidden human labor behind AI - Historical context and political analysis - Accessible writing style for complex topics Disliked: - Limited discussion of potential solutions - Some repetition across chapters - More critique than technical analysis - Focus on problems rather than benefits of AI Ratings: Goodreads: 4.3/5 (1,200+ ratings) Amazon: 4.5/5 (250+ ratings) Reader quote: "Crawford takes us to the lithium mines, data-labeling clickworks, and surveillance systems that power AI, showing the real costs hidden behind the tech industry's promises." - Goodreads reviewer Several technical readers noted the book serves better as a social critique than a technical resource.

Race After Technology by Ruha Benjamin This book examines how AI and technology systems perpetuate racial hierarchies and social inequalities through design choices and data practices.

Weapons of Math Destruction by Cathy O'Neil The book reveals how algorithmic decision-making systems in finance, education, and employment create feedback loops that harm marginalized communities.

The Age of Surveillance Capitalism by Shoshana Zuboff This work traces how tech companies extract and monetize personal data to predict and influence human behavior at scale.

New Dark Age by James Bridle The book explores how the proliferation of technology and computational systems leads to environmental degradation and societal opacity.

A Human's Guide to Machine Intelligence by Kartik Hosanagar This examination breaks down how algorithmic decision-making systems shape human choices across platforms and industries while considering the broader societal implications.

🔹 Kate Crawford's research for Atlas of AI took her to lithium mines in Nevada, Amazon warehouses, and Victorian museums, revealing the physical and environmental impact of AI systems most users never see. 🔹 The book challenges the common perception of AI as purely digital by highlighting how AI systems rely on massive amounts of natural resources, from rare earth minerals to water for cooling data centers. 🔹 Author Kate Crawford co-founded the AI Now Institute at New York University, the first university research center focused exclusively on the social implications of artificial intelligence. 🔹 Atlas of AI examines how AI systems often perpetuate historical inequalities, showing how 19th-century phrenology and colonial classification systems influence modern facial recognition technologies. 🔹 The book reveals that training a single large AI language model can emit as much carbon dioxide as five cars over their entire lifetimes.