Language Machines: Cultural AI and the End of Remainder Humanism
Sherman Kent, under whose mighty shadow analysts labor, decreed that intelligence analysis exists to serve policymakers, not to pursue knowledge for its own sake. As CIA tradecraft expert Jack Davis reminded us, Kent was pretty blunt about this—intelligence that gets ignored is "useless.” Kent made it look easy, his method, drawn from a background in historical scholarship, deceptively simple: test your sources, then check your own biases before drawing conclusions. As any analyst will tell you, the former is simple enough in theory; the latter is brutally difficult in practice.
This grounding matters when we turn to Leif Weatherby’s Language Machines, a likewise deceptively simple book that examines generative artificial intelligence (GenAI) through cultural theory rather than intelligence tradecraft. Weatherby uses structuralism and semiotics to make a striking claim—that large language models (LLMs) function as ideology machines, creating meaning through pattern recognition while embedding cultural assumptions at scale. For veterans of the intelligence trade, this isn’t just academic theorizing. It means, in a nutshell, that the bias problems Richards Heuer warned about have mutated into something a thousand times harder to spot and a million times easier to spread.