Kate Crawford: The scholar who redrew AI as a map of mines, labor, and power

Kate Crawford came to AI research from outside computer science entirely. In the late 1990s she was part of the Canberra electronic music duo B(if)tek, releasing three albums between 1998 and 2003, before completing a PhD at the University of Sydney and building an academic career studying how media technologies reorganize social life.

Her early research tracked mobile phones, social networks, and the cultures forming around them. In 2012 she co-authored Critical Questions for Big Data with danah boyd, one of the earliest widely cited papers to argue that large datasets were not objective observers of society but products of choices about what to collect and how to classify it. That paper established the posture she has held since: the politics are inside the dataset, not bolted on afterward.

She joined Microsoft Research, where she remains a senior principal researcher in New York, and holds a research professorship in communication and science and technology studies at USC Annenberg.

In 2017 Crawford co-founded the AI Now Institute at New York University with Meredith Whittaker, one of the first research organizations dedicated to the social implications of AI rather than its technical frontiers. AI Now's reports on surveillance, algorithmic management, and the lack of accountability in deployed systems helped move those topics from activist fringe to policy agenda.

In 2019 she became the inaugural holder of the AI and Justice visiting chair at the École Normale Supérieure in Paris. Over the following years she advised policymakers at the United Nations, the Federal Trade Commission, the European Parliament, and the White House. Her argument in those rooms has been consistent: ethics frameworks are necessary but not sufficient, and the sharper questions are who benefits, who is harmed, and whether a system concentrates power in already powerful institutions.

In 2018 Crawford and the artist Vladan Joler published Anatomy of an AI System, an enormous map and essay tracing a single Amazon Echo back through its full existence: the lithium mines and rare earth extraction, the assembly labor, the data annotation, the energy grid, and the device's eventual afterlife as electronic waste. The work won the Beazley Design of the Year award in 2019 and entered the permanent collections of the Museum of Modern Art in New York and the Victoria and Albert Museum in London.

That project became the spine of her 2021 book, Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence, published by Yale University Press. The book's chapters move from earth to labor to data to classification to power, arguing at each layer that what the industry calls intelligence is a vast, mostly hidden infrastructure of extraction and human work. It arrived months before ChatGPT and gave the coming debate much of its critical vocabulary.

Alongside the book, Crawford pursued the same argument through exhibitions. Training Humans, her 2019 collaboration with the artist Trevor Paglen at the Fondazione Prada in Milan, displayed the actual photographs used to train facial recognition systems, showing how people had been sorted into categories without their knowledge. The accompanying essay, Excavating AI, examined how benchmark datasets like ImageNet encoded the judgments of their makers.

The exhibition was also criticized, including by other scholars, for displaying facial images without the subjects' informed consent, a charge that cut close to the exhibition's own subject. Crawford engaged with the criticism rather than dismissing it, and the episode stands as a useful complication in her record: the politics of data apply to critics of data politics too.

Crawford's recent work has widened the lens again. Calculating Empires, a second large-scale collaboration with Joler completed in 2023, maps five centuries of technology and power back to 1500, placing AI in a genealogy of colonialism, classification, and control. The piece won the European Commission's S+T+ARTS Grand Prize in 2024, was shown at the Jeu de Paume in Paris in 2025, and received a Silver Lion at the 2025 Venice Architecture Biennale. She was named to the TIME100 AI list in 2023.

The honest assessment cuts in both directions. Her core claims have aged well: the energy and water demands of data centers, the hidden labor of annotation and moderation, and the political content of training data are now mainstream concerns priced into regulation and litigation. The contested ground is the frame itself. Critics argue that the extractive-industry reading understates what these systems deliver to users, and some note the tension of building a critique of concentrated AI power from inside one of its largest corporate laboratories. Crawford's answer, consistent across two decades, is that the material facts are the argument, and the facts are now very hard to avoid.