Playing Dumb: How a Chess Transformer Simulates Users Across Skill Levels
Organizations: Department of Computer Science, University of Toronto
Abstract
Generative models increasingly simulate users at different skill levels in education and AI evaluation. Yet it remains unclear whether lower-skill behavior reflects less internal knowledge or a different use of retained knowledge. This distinction is difficult to study in natural language, where proficiency and relevant knowledge are hard to measure. In this work, we study Maia-2, a chess transformer that predicts human moves conditioned on player skill, using quantitative skill ratings and deterministically labeled chess concepts. We hold board positions and model weights fixed and analyze positions where skill conditioning changes the prediction from a blunder to the best move. First, we measure encoded knowledge by training probes to predict the presence of 172 chess concepts from Maia-2's internal states. Probe accuracy remains nearly constant across skill conditions, and probes trained at one level transfer accurately to the others. To test whether stronger decisions can be recovered from this retained knowledge, we fine-tune only Maia-2's final output layer, raising best-move accuracy in the lowest skill band from 14.4% to 82.8%. Training this layer only on highest-skill representations retains most of the improvement at lower levels, showing that recovery does not require direct training on lower-skill states. Steering concept-relevant features learned by a sparse autoencoder also improves lower-skill decisions while leaving probe-measured concept encoding practically unchanged. These results support an account in which skill conditioning changes how retained concept knowledge influences move selection, showing that weaker behavior can understate the knowledge available within a model.
Figures & tables
| Baseline | Steering (Random) | Steering (Feature) | Head SFT | |
| Best-Move Accuracy | ||||
| 1200–1399 | ||||
| 1400–1599 | ||||
| 1600–1799 | ||||
| 1800–1999 | ||||
Appendix figures & tables10 assets
Supplementary material from the paper’s appendix.
Appendix
| Concept names | Description |
|---|---|
| Square Control — Active Player (64 concepts) | |
| offensive_threat_{sq} sq {a1, …, h8} | True if the opponent has a piece on square sq with strictly more active-player attackers than opponent defenders. Instantiated for each of the 64 squares. |
| Square Control — Opponent (64 concepts) | |
| defensive_threat_{sq} sq {a1, …, h8} | True if the active player has a piece on square sq with strictly more opponent attackers than own defenders. Instantiated for each of the 64 squares. |
| King and Queen Safety (4 concepts) | |
| king_danger [m | o] | True if 5 or more squares adjacent to the side’s king are attacked by the other side. |
| Candidate transitional positions | 2.28M |
| Blundered transitional positions | 1.15M |
| Training split | 0.80M |
| Validation split | 0.11M |
| Test split | 0.23M |
Transformer Interpretability
Transformer interpretability studies how attention, feed-forward layers, and hidden representations support model predictions and behavior.
See all 356 papers