AI as an object of study

Social science for artificial decision-making.

Artificial Experiments uses the full methodological tradition of the social sciences and humanities to understand artificial intelligence as a complex decision-making system.

Programme

Centuries of methods for a new form of intelligence.

Social sciences and humanities were built to study complex intelligence in context: preferences, institutions, incentives, interpretation, strategic interaction, language, culture, and power. Artificial Experiments brings that methodological repertoire to AI systems.

The work treats artificial intelligence as an empirical object. Conjoint experiments, list experiments, strategic games, textual analysis, and interpretive methods help make model behaviour measurable without reducing it to a single benchmark.

4working papers
3core method families
AIas experimental subject
Essays
Working papers