AI and social theory

Abstract In this paper, we sketch a programme for AI-driven social theory. We begin by defining what mean artificial intelligence (AI) in context. then lay out our specification how AI-based models can draw on the growing availability of digital data to help test validity different theories based their predictive power. doing so, use work Randall Collins and his state breakdown model exemplify that, already today, synthesise knowledge from variety sources, reason about world, apply is known across wide range problems systematic way. However, also find that theory remains subject practical, technical, epistemological limitations. Most critically, existing AI-systems lack three essential capabilities needed advance ways are cumulative, holistic, open-ended, purposeful. These (1) semanticisation, i.e., ability develop operationalize verbal concepts represent machine-manipulable knowledge; (2) transferability, transfer has been learned one context another; (3) generativity, independently create improve models. argue if gaps identified here addressed further research, there no why, future, most advanced should not be led cumulative advances.