Training with AI – Evidence from Chess Computers

Can training with artificial intelligence improve human decision-making in complex strategic interactions? This study examines whether chess computers – an early incarnation of artificial intelligence – have helped human players significantly improve their game. While artificial intelligence can increasingly perform complex cognitive tasks independently, its potential to train and augment human strategic capabilities has remained unexplored. Using novel data on over 20,000 chess players and granular information on over 500,000 tournament games, we provide compelling empirical evidence that access to chess computers in the 1970s and 1980s led to measurable performance improvements for human players. But human players do not really learn the same when they rely on artificial rather than human training partners.

Publication

Gaessler, Fabian; Piezunka, Henning (2023). Training with AI: Evidence from Chess Computers, Strategic Management Journal, 44 (11), 2724–2750, External Link (DOI)