Aligning Superhuman AI with Human Behavior: Chess as a Model System
Open accessThe DOI matches its Crossref record; 6 of the 6 references it cites were found in a registry or an index.
Abstract
As artificial intelligence becomes increasingly intelligent---in some cases, achieving superhuman performance---there is growing potential for humans to learn from and collaborate with algorithms. However, the ways in which AI systems approach problems are often different from the ways people do, and thus may be uninterpretable and hard to learn from. A crucial step in bridging this gap between human and artificial intelligence is modeling the granular actions that constitute human behavior, rather than simply matching aggregate human performance. We pursue this goal in a model system with a long history in artificial intelligence: chess. The aggregate performance of a chess player unfolds as they make decisions over the course of a game. The hundreds of millions of games played online by players at every skill level form a rich source of data in which these decisions, and their exact context, are recorded in minute detail. Applying existing chess engines to this data, including an open-source implementation of AlphaZero, we find that they do not predict human moves well. We develop and introduce Maia, a customized version of AlphaZero trained on human chess games, that predicts human moves at a much higher accuracy than existing engines, and can achieve maximum accuracy when predicting decisions made by players at a specific skill level in a tuneable way. For a dual task of predicting whether a human will make a large mistake on the next move, we develop a deep neural network that significantly outperforms competitive baselines. Taken together, our results suggest that there is substantial promise in designing artificial intelligence systems with human collaboration in mind by first accurately modeling granular human decision-making.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Mastering the game of Go without human knowledge
Silver, D., Schrittwieser, J., Simonyan, K., Antonoglou, I., Huang, A., Guez, A., Hubert, T., Baker, L., Lai, M., Bolton, A., et al. (2017) · Nature · doi:10.1038/nature24270 · Cited by 9,084
Lipton, Z. C. (2018) · Queue · doi:10.1145/3236386.3241340 · Cited by 5,427
A general reinforcement learning algorithm that masters chess, shogi, and Go through self-play
Silver, D., Hubert, T., Schrittwieser, J., Antonoglou, I., Lai, M., Guez, A., Lanctot, M., Sifre, L., Kumaran, D., Graepel, T., et al. (2018) · Science · doi:10.1126/science.aar6404 · Cited by 3,562 · cited 3×
“revolutionized algorithmic game-playing with the introduction of a sequence of deep reinforcement learning frameworks culminating in AlphaZero, an algorithm that achieved superhuman performance in chess, shogi, and Go by learning from self-play [14, 15].”
BackgroundMethodPrinciples of mixed-initiative user interfaces
Horvitz, E. (1999) · Proceedings of the SIGCHI conference on Human factors in computing systems the CHI is the limit - CHI '99 · doi:10.1145/302979.303030 · Cited by 1,490
Human Decisions and Machine Predictions
Kleinberg, J., Lakkaraju, H., Leskovec, J., Ludwig, J., & Mullainathan, S. (2018) · The Quarterly Journal of Economics · doi:10.1093/qje/qjx032 · Cited by 1,040
Self-citationInterpretable classification models for recidivism prediction
Zeng, J., Ustun, B., & Rudin, C. (2017) · Journal of the Royal Statistical Society Series A: Statistics in Society · doi:10.1111/rssa.12227 · Cited by 220
Cited by
Acquisition of chess knowledge in AlphaZero
McGrath, T., Kapishnikov, A., Tomašev, N., Pearce, A., Wattenberg, M., Hassabis, D., Kim, B., Paquet, U., & Kramnik, V. (2022) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.2206625119 · Cited by 166
“AI has also been used to analyze chess itself by modeling differences between the play style (38) or strength (39) of different human players as well as machine performance on chess variants (40–42).”
MethodHybrid social learning in human-algorithm cultural transmission
Brinkmann, L., Gezerli, D., Kleist, K. V., Müller, T. F., Rahwan, I., & Pescetelli, N. (2022) · Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences · doi:10.1098/rsta.2020.0426 · Cited by 24
Time spent thinking in online chess reflects the value of computation
Russek, E., Acosta-Kane, D., van Opheusden, B., Mattar, M. G., & Griffiths, T. (2022) · doi:10.31234/osf.io/8j9zx · Cited by 18
Kumar, H., Yoo, S., Bernuy, A., Shi, J., Luo, H., Williams, J. J., Kuzminykh, A., Anderson, A., & Kornfield, R. (2024) · doi:10.2139/ssrn.4892998 · Cited by 9
Self-citationSelf-Regulated AI Use Hinders Long-Term Learning
Poulidis, S., Bastani, H., & Bastani, O. (2025) · doi:10.2139/ssrn.5604932 · Cited by 1
Metadata
- Authors
- McIlroy-Young, R., Sen, S., Kleinberg, J., & Anderson, A.
- Venue
- Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining
- Publisher
- ACM
- Published
- 2020-08-20
- Pages
- 1677-1687
- OpenAlex
- W3033620876
- Cited by
- 99
- References
- 6
- FWCI
- 4.43
- Type
- conference paper
- Language
- en
- Open access
- bronze
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Aligning Superhuman AI with Human Behavior: Chess as a Model System. https://app.cochat.ai/papers/10.1145/3394486.3403219
- The paper
- McIlroy-Young, R., Sen, S., Kleinberg, J., & Anderson, A. (2020). Aligning Superhuman AI with Human Behavior: Chess as a Model System. Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. https://doi.org/10.1145/3394486.3403219
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.