The DOI matches its Crossref record; 9 of the 9 references it cites were found in a registry or an index.
Acquisition of chess knowledge in AlphaZero
Open accessAbstract
We analyze the knowledge acquired by AlphaZero, a neural network engine that learns chess solely by playing against itself yet becomes capable of outperforming human chess players. Although the system trains without access to human games or guidance, it appears to learn concepts analogous to those used by human chess players. We provide two lines of evidence. Linear probes applied to AlphaZero’s internal state enable us to quantify when and where such concepts are represented in the network. We also describe a behavioral analysis of opening play, including qualitative commentary by a former world chess champion.
Overview
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Probes have also been applied in vision and deep reinforcement learning.
“They have also been applied in vision (Alain & Bengio, 2017;Kim et al., 2018b) and deep reinforcement learning (McGrath et al., 2022;Forde et al., 2023).”
From Open Problems in Mechanistic Interpretability (2025) · cites this paper as McGrath et al., 2022
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
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474
A Unified Approach to Interpreting Model Predictions
Lundberg, S., & Lee, S.-I. (2017) · arXiv · doi:10.48550/arxiv.1705.07874 · Cited by 23,707
Barredo Arrieta, A., Díaz-Rodríguez, N., Del Ser, J., Bennetot, A., Tabik, S., Barbado, A., Garcia, S., Gil-Lopez, S., Molina, D., Benjamins, R., et al. (2020) · Information Fusion · doi:10.1016/j.inffus.2019.12.012 · Cited by 10,423
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 2×
“Progress in chess AI has followed the advent of modern computing, ranging from advances in symbolic AI to recent self-learned neural network engines, like AlphaZero (3) and Lc0 (32).”
BackgroundMethodSelf-citationTowards A Rigorous Science of Interpretable Machine Learning
Doshi-Velez, F., & Kim, B. (2017) · arXiv · doi:10.48550/arxiv.1702.08608 · Cited by 3,121
Self-citationHuman-Centered Tools for Coping with Imperfect Algorithms During Medical Decision-Making
Cai, C. J., Reif, E., Hegde, N., Hipp, J., Kim, B., Smilkov, D., Wattenberg, M., Viegas, F., Corrado, G. S., Stumpe, M. C., et al. (2019) · Proceedings of the 2019 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3290605.3300234 · Cited by 405
Self-citationZoom In: An Introduction to Circuits
Olah, C., Cammarata, N., Schubert, L., Goh, G., Petrov, M., & Carter, S. (2020) · Distill · doi:10.23915/distill.00024.001 · Cited by 288
Explainable Reinforcement Learning through a Causal Lens
Madumal, P., Miller, T., Sonenberg, L., & Vetere, F. (2020) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v34i03.5631 · Cited by 58
Flagged first, then by citations
Research works printed but not linked: 41
- A. Turing, “Digital computers applied to games” in Faster than Thought: A Symposium on Digital Computing Machines, B. V. Bowden, Ed. (Pitman Publishing, London, United Kingdom, 1953), pp. 286–310.
- 10.1016/S0004-3702(01)00129-1 · DOI · Paper page
- 10.1073/pnas.1907375117 · DOI · Paper page
- 10.23915/distill.00010 · DOI · Paper page
- B. Kim . “Interpretability beyond feature attribution: Quantitative testing with concept activation vectors (TCAV)” in International Conference on Machine Learning (PMLR 2018) pp. 2668–2677.
- 10.1073/pnas.1907367117 · DOI
- 10.23915/distill.00030 · DOI · Paper page
- D. Bau B. Zhou A. Khosla A. Oliva A. Torralba “Network dissection: Quantifying interpretability of deep visual representations” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (2017) pp. 6541–6549. · DOI · Paper page
- D. Alvarez-Melis, T. S. Jaakkola, “Towards robust interpretability with self-explaining neural networks” in Advances in Neural Information Processing Systems (2018), pp. 7786–7795.
- A. Ghorbani, J. Wexler, J. Y. Zou, B. Kim, “Towards automatic concept-based explanations” in Advances in Neural Information Processing Systems (2019), pp. 9273–9282.
- P. W. Koh . Concept bottleneck models. arXiv [Preprint] (2020). https://arxiv.org/abs/2007.04612 (Accessed 3 November 2022). · arXiv
- M. T. Ribeiro S. Singh C. Guestrin “Why should I trust you?: Explaining the predictions of any classifier” in Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (ACM 2016) pp. 1135–1144. · DOI · Paper page
- M. Sundararajan A. Taly Q. Yan “Axiomatic attribution for deep networks” in Proceedings of the 34th International Conference on Machine Learning (2017) vol. 70 pp. 3319–3328.
- J. R. Clough . “Global and local interpretability for cardiac MRI classification” in International Conference on Medical Image Computing and Computer-Assisted Intervention (Springer 2019) pp. 656–664. · DOI · Paper page
- M. Graziani V. Andrearczyk H. Müller “Regression concept vectors for bidirectional explanations in histopathology” in Understanding and Interpreting Machine Learning in Medical Image Computing Applications (Springer 2018) pp. 124–132. · DOI
- D. Mincu . “Concept-based model explanations for electronic health records” in Proceedings of the Conference on Health Inference and Learning (2021) pp. 36–46. · DOI
- D. Bouchacourt L. Denoyer EDUCE: Explaining model Decisions through Unsupervised Concepts Extraction. arXiv [Preprint] (2019). https://arxiv.org/abs/1905.11852 (Accessed 3 November 2022). · arXiv
- H. Yeche J. Harrison T. Berthier “Ubs: A dimension-agnostic metric for concept vector interpretability applied to radiomics” in Interpretability of Machine Intelligence in Medical Image Computing and Multimodal Learning for Clinical Decision Support (Springer 2019) pp. 12–20. · DOI
- S. Sreedharan U. Soni M. Verma S. Srivastava S. Kambhampati Bridging the gap: Providing post-hoc symbolic explanations for sequential decision-making problems with black box simulators. arXiv [Preprint] (2020). https://arxiv.org/abs/2002.01080v2 (Accessed 3 November 2022). · arXiv
- G. Schwalbe M. Schels “Concept enforcement and modularization as methods for the ISO 26262 safety argumentation of neural networks” in Proceedings of the 10th European Congress on Embedded Real Time Software and Systems (Toulouse France 2020). · DOI · Paper page
- J. Z. Forde C. Lovering G. Konidaris E. Pavlick M. L. Littman “Where when & which concepts does AlphaZero learn? Lessons from the game of Hex” in AAAI Workshop on Reinforcement Learning in Games (2022) vol. 2.
- N. Tomlin A. He D. Klein Understanding game-playing agents with natural language annotations. arXiv [Preprint] (2022). https://arxiv.org/abs/2204.07531 (Accessed 3 November 2022). · arXiv
- Y. Goyal A. Feder U. Shalit B. Kim Explaining classifiers with causal concept effect (CaCE). arXiv [Preprint] (2019). https://arxiv.org/abs/1907.07165v1 (Accessed 3 November 2022). · arXiv
- M. T. Bahadori D. E. Heckerman Debiasing concept bottleneck models with instrumental variables. arXiv [Preprint] (2020). https://arxiv.org/abs/2007.11500v2 (Accessed 3 November 2022). · arXiv
- 10.1007/3-540-60364-6_40 · DOI
- P. Gupta . “Explain your move: Understanding agent actions using specific and relevant feature attribution” in International Conference on Learning Representations (ICLR) (2020).
- H. Jhamtani V. Gangal E. Hovy G. Neubig T. Berg-Kirkpatrick “Learning to generate move-by-move commentary for chess games from large-scale social forum data” in The 56th Annual Meeting of the Association for Computational Linguistics (ACL) (2018). · DOI
- V. Cirik L. P. Morency E. Hovy “Chess Q&A: Question answering on chess games” in Reasoning Attention Memory (RAM) Workshop (Neural Information Processing Systems 2015).
- I. Kamlish, I. B. Chocron, N. McCarthy, M. A. T. E. Senti, Learning to play chess through natural language processing. arXiv [Preprint] (2019). https://arxiv.org/abs/1907.08321 (Accessed 3 November 2022). · arXiv
- R. McIlroy-Young, R. Wang, S. Sen, J. Kleinberg, A. Anderson, Learning personalized models of human behavior in chess. arXiv [Preprint] (2020). https://arxiv.org/abs/2008.10086v1 (Accessed 3 November 2022). · arXiv
- R. McIlroy-Young S. Sen J. Kleinberg A. Anderson “Aligning superhuman AI with human behavior: Chess as a model system” in Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining (Association for Computing Machinery New York NY 2020) pp. 1677–1687. · DOI
- 10.3389/frai.2020.00024 · DOI
- N. Tomašev, U. Paquet, D. Hassabis, V. Kramnik, Assessing game balance with AlphaZero: Exploring alternative rule sets in chess. arXiv [Preprint] (2020). https://arxiv.org/abs/2009.04374 (Accessed 3 November 2022). · arXiv
- 10.1145/3460349 · DOI
- T. McGrath . Acquisition of chess knowledge in AlphaZero. arXiv [Preprint] (2021). https://arxiv.org/abs/2111.09259v2 (Accessed 3 November 2022). · arXiv
- M. Sadler, N. Regan, Game Changer: AlphaZero’s Groundbreaking Chess Strategies and the Promise of AI (New In Chess, 2019).
- 10.1162/coli_a_00422 · DOI
- Y. Xu S. Zhao J. Song R. Stewart S. Ermon “A theory of usable information under computational constraints” in International Conference on Learning Representations (2019).
- J. MacGlashan M. L. Littman “Between imitation and intention learning” in Twenty-Fourth International Joint Conference on Artificial Intelligence (2015).
- 10.1038/44565 · DOI
- 10.23915/distill.00029 · DOI · Paper page
Cited by
Sorted by citations
Open Problems in Mechanistic Interpretability
Sharkey, L., Chughtai, B., Batson, J., Lindsey, J., Wu, J., Bushnaq, L., Goldowsky-Dill, N., Heimersheim, S., Ortega, A., Bloom, J., et al. (2025) · arXiv · doi:10.48550/arxiv.2501.16496 · Cited by 291
Self-citationThe alignment problem from a deep learning perspective
Ngo, R., Chan, L., & Mindermann, S. (2022) · arXiv · doi:10.48550/arxiv.2209.00626 · Cited by 68
“…policies learned, one example of a model-based policy learning robust outcome representations comes from AlphaZero, which learned a range of human chess concepts, including concepts used in top chess engine Stockfish’s hand-crafted evaluation function (e.g. “king safety”) [McGrath et al., 2021].”
BackgroundEmergent Linear Representations in World Models of Self-Supervised Sequence Models
Nanda, N., Lee, A., & Wattenberg, M. (2023) · Proceedings of the 6th BlackboxNLP Workshop: Analyzing and Interpreting Neural Networks for NLP · doi:10.18653/v1/2023.blackboxnlp-1.2 · Cited by 55
Self-citationRobust agents learn causal world models
Richens, J., & Everitt, T. (2024) · arXiv · doi:10.48550/arxiv.2402.10877 · Cited by 5
“Several recent empirical works have explored if deep learning models learn ‘surface statistics’ (e.g. correlations between inputs and outputs) or learn internal representations of the world (McGrath et al., 2022; Abdou et al., 2021; Li et al., 2022; Gurnee & Tegmark, 2023).”
Background
Metadata
- Authors
- McGrath, T., Kapishnikov, A., Tomašev, N., Pearce, A., Wattenberg, M., Hassabis, D., Kim, B., Paquet, U., & Kramnik, V.
- Affiliations
- Google DeepMind (United Kingdom) (GB)
- Google (United States) (US)
- World Chess Champion, 2000–2007
- Corresponding author
- McGrath, Thomas
- ORCID
- Venue
- Proceedings of the National Academy of Sciences
- Publisher
- National Academy of Sciences
- Published
- 2022-11-14
- Volume
- 119
- Issue
- 47
- Pages
- e2206625119-e2206625119
- ISSN
- 0027-8424 (ISSN-L)
- 1091-6490
- PMC
- PMC9704706
- OpenAlex
- W3212876771
- Cited by
- 166
- References
- 9
- Citations by year
- 2026: 29 · 2025: 25 · 2024: 38 · 2023: 41 · 2022: 33
- FWCI
- 9.17
- Type
- journal article
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- Locations
- Proceedings of the National Academy of Sciences (publishedVersion, cc-by)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Artificial Intelligence in Games
- Physical Sciences › Computer Science › Artificial Intelligence › Artificial Intelligence Applications
- Physical Sciences › Computer Science › Artificial Intelligence › Reinforcement Learning in Robotics
- Keywords
- AlphaZero, chess, knowledge acquisition, self-play reinforcement learning, linear probing, internal representations, behavior analysis, neural network interpretability
- MeSH terms
- Humans, Learning, Recreation, Neural Networks, Computer
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
