No. 35 of 500 No. 11 in AI Ethics & Society

Atlas of AI

Power, Politics, and the Planetary Costs of Artificial Intelligence

By Kate Crawford

Published
August 16, 2022
Print length
336 pages
Reading level
Intermediate
ISBN-13
978-0300264630
Publisher
Yale University Press
Written for
Policy professionals, students and executives who want to understand AI's environmental, labor and political footprint.

Synopsis

Kate Crawford sets out to map artificial intelligence as a physical and political system rather than as disembodied code. Each chapter visits a different layer of what it takes to build AI. "Earth" starts in Nevada's lithium country and follows the minerals and energy behind computing; "Labor" examines the crowdworkers who label data and the warehouse staff managed by algorithms; "Data" and "Classification" show how training sets are scraped without consent and how their categories carry old biases into new systems. "Affect" questions the science behind emotion-recognition tools, and "State" examines the ties between AI companies, intelligence agencies and the military.

Crawford's central claim is that AI is an extractive industry that draws on the planet's resources, low-paid human effort and personal data, and that it tends to concentrate power in the institutions that already hold it. The book rests on a decade of research, site visits and archival work, and its prose suits general readers as well as students. Some reviewers found the chapters overlapping and the argument rhetorical in places. Readers gain a framework for asking what any AI system costs, who pays and who decides, questions now central to debates over data centers, training data and surveillance.

Scorecard

Editorial score

How it scored

Editorial score: 77.6 / 1005 weighted criteria

  • Influence30%

    25.5/30
  • Authority25%

    21.3/25
  • Practical value20%

    12.0/20
  • Currency15%

    11.3/15
  • Readability10%

    7.5/10
  • Editorial score

    77.6/100

How the rank was computed

Rank No. 35 of 500 comes from a composite of 71.1: observable signals 36.1 of 55 points plus editorial review 35.0 of 45 points.

All 11 inputs
InputPoints
Observable signals36.1/55
Reach (Amazon ratings volume)10.1/16.5
Consensus across best-of lists13.8/13.8
Awards and bestseller evidence5.0/8.3
Reader rating, sample-adjusted5.4/8.3
Amazon sales rank1.8/5.5
Recency0.0/2.8
Editorial review35.0/45
Author authority9.6/11.3
Influence11.5/13.5
Practical value5.4/9.0
Readability3.4/4.5
Currency5.1/6.8

Each criterion is scored from 1 to 10 and weighted, and the points add up to the editorial score. The rank combines observable signals (55 percent) with this score (45 percent). Read the methodology.

Verdict

A rigorous, award-winning account of AI as an extractive industry, essential for understanding the material and political costs behind the software.

Strengths

  • Original material framing of AI's costs
  • Vivid field reporting and archival research
  • Widely cited across law and policy

Keep in mind

Readers looking for technical explanation or policy prescriptions will find neither, and some chapters repeat arguments made earlier in the book.

Key takeaways

  1. AI depends on mining, energy and water, so its environmental footprint begins long before any model is trained or deployed.
  2. Hidden human labor, from crowdworkers labeling data to algorithmically managed warehouse staff, sustains systems that are marketed as automatic.
  3. Training datasets are often scraped without consent, and their classification schemes can embed contested assumptions about race, gender and identity.
  4. Emotion-recognition tools rest on disputed science about facial expressions, yet vendors sell them for hiring, policing and security decisions.

Notable facts

  • Won the Society for the History of Technology's 2022 Sally Hacker Prize, whose committee called it "lucid in writing and argument." Source (opens in a new tab)
  • Yale University Press lists it as a 2021 Choice Outstanding Academic Title and winner of the 2022 ASIS&T Best Information Science Book award. Source (opens in a new tab)
  • The Financial Times named it one of the best books on technology of 2021. Source (opens in a new tab)

About the author

Kate Crawford

Kate Crawford is a research professor at USC Annenberg and a senior principal researcher at Microsoft Research in New York. She co-founded the AI Now Institute at NYU in 2017 and held the inaugural AI and Justice chair at the École Normale Supérieure in Paris. Her map Anatomy of an AI System, made with Vladan Joler, is held in the collections of MoMA and the V&A.

Questions

What is Atlas of AI about?
Atlas of AI treats artificial intelligence as a physical and political system. Kate Crawford follows its supply chain from lithium mines and energy use through crowdworker labor, scraped training data, flawed classification schemes and emotion-recognition tools to its use by states and militaries. Her argument is that AI is an extractive industry that tends to concentrate power among those who already hold it.
Is Atlas of AI still relevant in 2026?
Largely, yes. The book predates ChatGPT, so it does not cover large language models directly, but its core concerns have grown more pressing: data center energy and water use, the human labor behind model training, disputes over scraped data, and AI's role in surveillance. Readers will need newer books for the details of generative AI.
Who should read Atlas of AI?
It suits policymakers, students in technology and society courses, journalists and executives who want to understand costs that never appear in a model's benchmark scores. No technical background is required. Readers who want coding guidance, business strategy or detailed regulatory proposals should choose another book, since Crawford diagnoses problems more than she prescribes fixes.

Sources

As an Amazon Associate, Best AI Books earns from qualifying purchases. Best AI Books is sponsored by Iternal.ai, whose founder and CEO, John Byron Hanby IV, wrote The AI Strategy Blueprint, our #1 ranked book. How we rank books.