Critical analyses of AI usually adopt a stance of defensive humanism. Instead, Nan Da interrogates its mode of reasoning. How do LLMs make the step from data to inference—and what does it mean when ...