Traces how generative AI folds into existing patterns of prejudice, propaganda and political extremities

From deep fakes and voice clones to synthetic media, the rise of artificial intelligence is often cast as a monumental crisis—a "democracy-killing machine" capable of unravelling the social fabric. In Divided by Machines, Sahana Udupa offers a vital intervention, arguing that generative AI does not operate in a vacuum. Instead, it folds into existing paradigms of political disinformation and long-standing patterns of online and offline prejudice.

Udupa explores the tension between AI as a hypervisible threat and its role as generic infrastructure, normalized through the recommendation algorithms and voice assistants that shape our daily lives. Drawing on diverse ethnographic research—ranging from right-wing discourses and Elon Musk’s theatrics to deep fakes in Germany, misogyny in Brazil, and digital influence operations in India—she reveals how AI technologies are woven into social scenes of vitriol. She proposes “AI extremities” as sociocultural constellations of synthetic speech, deception and entrenched practices of exclusion and belonging. By centering these sociocultural constellations, Udupa demonstrates that social worlds do not merely contextualize technology; they define its limits and determine its power.

Moving beyond the binary of dystopian dread and techno-solutionism, Divided by Machines provides a nuanced look at AI extremities while dismantling the myth of technological inevitability. It offers a necessary framework for reimagining a digital future that respects responsible barriers in the political sphere. Ultimately, Udupa advocates for a pivotal principle of slowing down.