A lab for studying artificial decision-making.

Artificial Experiments studies AI systems as objects of inquiry. The central aim is to use the full methodology of the social sciences and humanities to understand how artificial intelligence evaluates choices, reacts to context, reasons strategically, and reveals or conceals preferences.

The initiative draws on centuries of work on complex intelligence: survey experiments, conjoint analysis, list experiments, game theory, institutional analysis, textual interpretation, historical reasoning, philosophy, and close reading. These methods offer a vocabulary for studying decision-making systems when the underlying mechanism is not fully observable.

Maxim Chupilkin is a PhD researcher at the University of Oxford working on AI-related research at the intersection of political science, economics, and model evaluation.

His current work uses experimental methods to study AI decision-making in military, economic, strategic, and sensitive-belief settings.

Artificial Experiments is an independent research initiative. Views expressed here are personal and do not represent any institution with which Maxim Chupilkin is affiliated.