Kate Crawford. Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence. New Haven: Yale University Press, 2022.
REVIEWED BY MARK DOUGLAS, PASTCF VICE PRESIDENT
In the days before the internet, I sometimes needed a map to drive to an unfamiliar address in town (especially during my brief stint delivering pizzas). But to drive longer distances, I needed an atlas — a book of maps that not only clarified connections between various geographies but also provided depth or texture to space by providing city inlays and other information relevant to my journeys.

That Kate Crawford names her book an “Atlas” rather than a “Map” is, therefore, clarifying. She isn’t simply mapping out the terrain described by, e.g., Chat-GPT and OpenAI; she’s inviting us to familiarize ourselves with the various maps that connect AI to the wider world and to gain a greater sense of the depths and complexities of that world as it shapes and is shaped by AI. She begins, therefore, not in the hyperventilating style of those fixated on threats from impending robot overlords but, instead, in the fundamental question, “How is intelligence ‘made,’ and what traps can that create?” (4), especially as human beings have repeatedly shown an ability to map notions of intelligence by reference to human minds. Such anthropomorphizing, Crawford argues, conceals many of the most significant moral, environmental, and technological problems with contemporary work in computer technology generally and AI specifically.
She begins by describing the enormous and deleterious effects on the natural world that come not only with the mining of rare earth metals and lithium but with the tech industry’s voracious demand for fossil fuels used to power and cool its systems. In an age increasingly shaped by climate change and environmental degradation, such demands not only come at huge costs; they come at costs that are hidden from users even as those costs are outsourced to vulnerable persons, communities, and the non-human natural world. From there, Crawford explores the costs of AI on human workers—most of whom are hidden from tech users and many of whom, again, are notably vulnerable to economic forces that micromanage their time and inhibit their possibilities for collective organizing and action. Pull back the curtain of artificial intelligence and we will discover rampant abuses of real people and real places.
AI is neither artificial nor intelligent
Having named the mining of materials and labor, Crawford explores the way technology mines data about us—generally without our awareness or approval—classifies that data in ways that replicate the racial, gender, and class biases of designers, deploys that data in support of algorithms that lack scientific credibility, all the while growing through its attachment to the military and state apparatuses of surveillance. By the end of the book, Crawford has provided voluminous evidence to substantiate her claim that “AI is neither artificial nor intelligent,” but is instead “designed to serve existing dominant interests. In this sense, artificial intelligence is a registry of power.” (8, emphasis hers).
The book provides an important and accessible set of maps that reveal not only the actual power of AI, but the real dangers that our fascination with technology may hide from us.








