Can AI Bluff and Detect Bluffs in Poker

Archived study

Records a simplified poker experiment with preset opponent bluff frequencies and saved model actions and confidence scores. The archived report compares calling and raising behavior across the three opponent groups.

Interpretation and limits

This small artificial river-decision task does not establish general poker strength or reliable bluff detection. Sparse bluff cases and the simplified environment limit the reported comparisons.

Package and provenance

The original .science export and matching poster are preserved byte-for-byte from the upstream case. The package contains the conversation, session metadata, and the resources inventoried in its internal manifest. This introduction summarizes the archived record; the research, external sources, and model evaluations were not rerun for this import.