Weird A.I.

Abstract

This essay sketches out what I call “weird AI”: probabilistic, associative systems that behave as if they feel. It shows how midcentury architects of artificial neural networks deliberately courted the uncanny as they engineered space for intuition, emotion, and nonrational thought, even as standard histories cast computing as an Enlightenment project of calculation and control. To make sense of artificial intelligence’s past and present, historians must move beyond an information-centric framework and reckon with the affective undercurrents that have shaped the field from its start.