The DOI matches its Crossref record; 30 of the 30 references it cites were found in a registry or an index; 5 of the 16 sources its claims cite were checked for support.
A general reinforcement learning algorithm that masters chess, shogi, and Go through self-play
Open accessAbstract
One program to rule them all: Computers can beat humans at increasingly complex games, including chess and Go. However, these programs are typically constructed for a particular game, exploiting its properties, such as the symmetries of the board on which it is played. Silver et al. developed a program called AlphaZero, which taught itself to play Go, chess, and shogi (a Japanese version of chess) (see the Editorial, and the Perspective by Campbell). AlphaZero managed to beat state-of-the-art programs specializing in these three games. The ability of AlphaZero to adapt to various game rules is a notable step toward achieving a general game-playing system. Science , this issue p. 1140 ; see also pp. 1087 and 1118
Overview
CoChat Summary
TL;DR
AlphaZero is a single reinforcement-learning and MCTS algorithm that, given only game rules, learned superhuman play in chess, shogi, and Go through self-play. It defeated world-champion programs Stockfish, Elmo, and AlphaGo Zero using the same algorithm and network architecture across all three games.
Key findings
- In chess, AlphaZero defeated Stockfish 8, winning 155 games and losing 6 out of 1,000; it also won matches from common human openings and 2016 TCEC opening positions.
- In shogi, AlphaZero defeated Elmo, winning 98.2% of games as black and 91.2% overall; in Go, it defeated AlphaGo Zero, winning 61% of games.
- AlphaZero first outperformed Stockfish after 4 hours (300,000 steps), Elmo after 2 hours (110,000 steps), and AlphaGo Lee after 30 hours (74,000 steps); training ran for 700,000 steps, taking about 9 hours in chess, 12 hours in shogi, and 13 days in Go.
- AlphaZero searched only about 60,000 positions per second in chess and shogi, versus 60 million for Stockfish and 25 million for Elmo, yet still defeated Stockfish with 1/10 thinking time and won 46% against Elmo with 1/100 thinking time.
Methods
- AlphaZero uses a deep neural network \(f_\theta(s)\) that outputs move probabilities \(p\) and a scalar value \(v\), trained by reinforcement learning from self-play starting from random parameters; MCTS uses these outputs to guide search, with 800 simulations per move during training.
- The same algorithm, network architecture, and hyperparameters were used for chess, shogi, and Go, with no domain knowledge except game rules; no opening book, endgame tablebases, or domain-specific heuristics were used.
- Training used 5,000 first-generation TPUs to generate self-play games and 16 second-generation TPUs to train networks; evaluation matches used 3 hours per game plus 15 seconds per move, with opponents on 44 CPU cores and AlphaZero on four first-generation TPUs and 44 CPU cores.
Limitations
- Stated: AlphaZero did not use opening books, endgame tablebases, or domain-specific heuristics; the authors note that some such techniques could further improve performance but were left for future research.
- Stated: hyperparameters were reused across games without game-specific tuning, except for exploration noise and the learning rate schedule; matches against an opening book were low in diversity, and forced randomization increased losses.
- Evident from the text: results cover only three games with separate trained instances, no cross-game transfer is demonstrated, and evaluation compares AlphaZero on TPU hardware against CPU-based opponents, so hardware and time-control differences qualify direct strength comparisons.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
3 Valid · 2 Not valid · 8 Not checked · 43 Own results and interpretations
Every source the paper cites for this claim was checked and supports it.
In the paper
“Apart from the representation of positions and actions described above, AlphaZero uses the same network architecture as AlphaGo Zero ( 9 ), briefly recapitulated here.”
What the sources say
- Supportedcites [9] source
The Neural Network Architecture Comparison (p. 29) identifies AlphaGo Zero’s architecture as a residual tower with policy and value heads. The citation supplies the AlphaGo Zero architecture that the citing authors say AlphaZero uses; the comparison to AlphaZero is their own statement. (LLM, conf=0.88)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“and in Go, AlphaZero first outperformed AlphaGo Lee ( 9 ) after 30 hours (74,000 steps).”
What the sources say
- Partly supportedcites [9] sourceExperimental — accuracy not yet measured
Section 2, p. 7 reports that AlphaGo Zero outperformed AlphaGo Lee, but says this happened after 36 hours, not 30. The passage does not give 74,000 steps, and it names the system AlphaGo Zero rather than AlphaZero. (LLM, conf=0.90)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The training algorithm achieved similar performance in all independent runs (see fig. S3), suggesting that the high performance of AlphaZero's training algorithm is repeatable.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“AlphaZero also defeated Stockfish when given 1 / 10 as much thinking time as its opponent (i.e. searching ~ 1 / 10 , 000 as many positions),”
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“and won 46% of games against Elmo when given 1 / 100 as much time (i.e. searching ~ 1 / 40 , 000 as many positions), see Fig. 2.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
23 more claims
Create a free account to see all 56 claims and to verify the 3 references still open.
- Probably not a substantial claim. 10
- It may state two separate claims as one. 23
- The quote only partly states this claim. 1
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?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Games have historically served as highly effective metrics for evaluating progress in deep reinforcement learning research”
From BALROG: Benchmarking Agentic LLM and VLM Reasoning On Games (2024) · cites this paper as Silver et al., 2018
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“In 2017, AlphaGo, using the same learning algorithms in TD-Gammon and a billion times more computing power, soundly beat Ke Jie, the world Go champion, in a game that is to chess what chess is to backgammon in complexity (Silver et al., 2018).”
From Large Language Models and the Reverse Turing Test (2023) · cites this paper as Silver et al., 2018
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Most of the attention was on who won, but more important was how it won: AlphaGo was creative and had discovered new moves and board positions that were outside human experience.”
From Large Language Models and the Reverse Turing Test (2023) · cites this paper as Silver et al., 2018
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“The primary method for applying more computation at inference, common in the AlphaZero family of agents2,72, is scaling the tree search and running multiple independent search attempts.”
From Olympiad-level formal mathematical reasoning with reinforcement learning (2026) · cites this paper as [2]
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“Modern AI can write computer programs (Li et al., 2022) or play chess at super-human level (Silver et al., 2018), but the ability of AI to perceive and act in the world remains far below human level.”
From Scaling Instructable Agents Across Many Simulated Worlds (2024) · cites this paper as Silver et al., 2018
1 more claim about this paper
Create a free account to see every claim other papers make about this one.
Citations
Cites
Mastering the game of Go with deep neural networks and tree search
Silver, D., Huang, A., Maddison, C. J., Guez, A., Sifre, L., van den Driessche, G., Schrittwieser, J., Antonoglou, I., Panneershelvam, V., Lanctot, M., et al. (2016) · Nature · doi:10.1038/nature16961 · Cited by 15,612
Self-citationAdvances in neural information processing systems 7
(1996) · Computers & Mathematics with Applications · doi:10.1016/s0898-1221(96)90463-0 · Cited by 14,281
Identity mappings in deep residual networks
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-46493-0_38 · Cited by 10,741
Identity Mappings in Deep Residual Networks
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · arXiv · doi:10.48550/arxiv.1603.05027 · Cited by 10,665
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
Self-citationIn-Datacenter Performance Analysis of a Tensor Processing Unit
Jouppi, N. P., Young, C., Patil, N., Patterson, D., Agrawal, G., Bajwa, R., Bates, S., Bhatia, S., Boden, N., Borchers, A., et al. (2017) · Proceedings of the 44th Annual International Symposium on Computer Architecture · doi:10.1145/3079856.3080246 · Cited by 4,537
Encyclopedia of Artificial Intelligence
(2009) · IGI Global eBooks · doi:10.4018/978-1-59904-849-9 · Cited by 1,396
Campbell, M., Hoane, A. J., & Hsu, F.-H. (2002) · Artificial Intelligence · doi:10.1016/s0004-3702(01)00129-1 · Cited by 1,226
Lim, T.-S., Loh, W.-Y., & Shih, Y.-S. (2000) · Machine Learning · doi:10.1023/a:1007608224229 · Cited by 1,133
An analysis of alpha-beta pruning
Knuth, D. E., & Moore, R. W. (1975) · Artificial Intelligence · doi:10.1016/0004-3702(75)90019-3 · Cited by 1,004
Research works printed but not linked: 7
- C. J. Maddison A. Huang I. Sutskever D. Silver paper presented at the International Conference on Learning Representations 2015 San Diego CA 7 to 9 May 2015.
- D. N. L. Levy M. Newborn How Computers Play Chess (Ishi Press 2009).
- W. Steinitz The Modern Chess Instructor (Edition Olms 1990).
- O. Arenz “Monte Carlo chess ” master’s thesis Technische Universität Darmstadt (2012).
- 10.1023/a:1007634325138 · DOI
- T. Kaneko K. Hoki in Advances in Computer Games – 13th International Conference ACG 2011 Revised Selected Papers Tilburg Netherlands 20 to 22 November 2011 (Springer 2012) pp. 158–169.
- R. Ramanujan A. Sabharwal B. Selman in Proceedings of the 26th Conference on Uncertainty in Artificial Intelligence (UAI 2010) Catalina Island CA 8 to 11 July (AUAI Press 2010).
Cited by
Items 801–900 of 3,622
Learning Models of Individual Behavior in Chess
McIlroy-Young, R., Wang, R., Sen, S., Kleinberg, J. M., & Anderson, A. (2022) · ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) · doi:10.1145/3534678.3539367 · Cited by 70
“The first, Leela Chess Zero , is an open source implementation of the deep reinforcement learning system AlphaZero Chess [58], and The second, Maia Chess , is a supervised learning adaptation of”
BackgroundOrorbia, M. E., & Warn, G. P. (2023) · Journal of Mechanical Design · doi:10.1115/1.4056693 · Cited by 21
Reinforcement Learning for the Face Support Pressure of Tunnel Boring Machines
Soranzo, E., Guardiani, C., & Wu, W. (2023) · Geosciences · doi:10.3390/geosciences13030082 · Cited by 20
“Although the study of reinforcement learning is still in its early stages, there have been some remarkable advancements, particularly those demonstrated by Google’s DeepMind research team [49–53], including the ability to excel at playing Atari video games and board games, such as chess and go.”
BackgroundMethodZhao, C., Yuan, S., Jiang, C., Cai, J., Yu, H., Wang, M. Y., & Chen, Q. (2023) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2023.3265893 · Cited by 24
“Reinforcement Learning for Manipulation: Reinforcement learning combined with deep learning has recently made extensive progress in learning skills in different domains, such as beyond human experts at the games of Go [21] and Atari [22].”
Probe microscopy is all you need *
Kalinin, S. V., Vasudevan, R. K., Liu, Y., Ghosh, A., Roccapriore, K. M., & Ziatdinov, M. A. (2023) · Machine Learning Science and Technology · doi:10.1088/2632-2153/acccd5 · Cited by 33
“For example, the success of AlphaGo [21, 125] or any other reinforcement learning algorithm for computer games is predicated on known rules and lack of external variables (e.g. the number of squares on the board will never change).”
BackgroundMethodGenerative Model-Based Testing on Decision-Making Policies
Li, Z., Wu, X., Zhu, D., Cheng, M., Chen, S., Zhang, F., Xie, X., Ma, L., & Zhao, J. (2023) · IEEE/ACM International Conference on Automated Software Engineering (ASE) · doi:10.1109/ase56229.2023.00153 · Cited by 23
“Although these methods have demonstrated effectiveness in various applications such as gaming [1], [2], Go [3], and robot manipulation [4], they may also exhibit erroneous behaviors leading to potential errors [5].”
MethodFu, M. C. (2019) · Asia Pacific Journal of Operational Research · doi:10.1142/s0217595919400098 · Cited by 20
Automatic discovery of interpretable planning strategies
Skirzyński, J., Becker, F., & Lieder, F. (2021) · Machine Learning · doi:10.1007/s10994-021-05963-2 · Cited by 21
“They were learned similarly as in AlphaZero (Silver et al., 2018), through Q function optimization and self-play, and by additionally learning the dynamics model.”
BackgroundMethodMAGIC: Learning Macro-Actions for Online POMDP Planning
Lee, Y., Cai, P., & Hsu, D. (2021) · Robotics: Science and Systems · doi:10.15607/rss.2021.xvii.041 · Cited by 34
“One can also bias action selection using a learned prior policy function [2, 41, 42, 43].”
BackgroundProspective Artificial Intelligence Approaches for Active Cyber DefencePreprint
Dhir, N., Hoeltgebaum, H., Adams, N. M., Briers, M., Burke, A., & Jones, P. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2104.09981 · Cited by 46
General Board Game ConceptsPreprint
Piette, É., Stephenson, M., Soemers, D. J. N. J., et al. (2021) · arXiv (Cornell University) · Cited by 22
Rezaei, A., Abdellatif, I., & Umar, A. (2025) · International Journal of Financial Studies · doi:10.3390/ijfs13010028 · Cited by 20
He, Z., Tian, W., Wang, J., Yan, H., Xin, K., & Tao, T. (2025) · Water Research · doi:10.1016/j.watres.2025.123954 · Cited by 23
Salas, J., Vidal, F. D. B., & Martinez-Trinidad, F. (2019) · IEEE Latin America Transactions · doi:10.1109/tla.2019.9011537 · Cited by 20
“do câncer [7], jogos eletrônicos [8], detecção de objetos [9], recomendação de vı́deo por contexto [10], reconhecimento de fala [11], análise de redes sociais [12], reconhecimento de ação em vı́deos [13] e mineração de texto [14].”
BackgroundSelf-Play Reinforcement Learning for Fast Image Retargeting
Kajiura, N., Kosugi, S., Wang, X., et al. (2020) · Tokyo Tech Research Repository (Tokyo Institute of Technology) · Cited by 19
Mean Field Games Flock! The Reinforcement Learning Way
Perrin, S., Laurière, M., Pérolat, J., Geist, M., Élie, R., & Pietquin, O. (2021) · International Joint Conference on Artificial Intelligence · doi:10.24963/ijcai.2021/50 · Cited by 20
“Combined with deep neural nets, RL has been used successfully to tackle problems which are too complex to be solved by exact methods [Silver et al. , 2018] or to address learning in multi-agent systems [Lanctot et al. , 2017].”
MethodTransferable learning on analog hardware
Vadlamani, S. K., Englund, D. R., & Hamerly, R. (2023) · Science Advances · doi:10.1126/sciadv.adh3436 · Cited by 29
Bousdekis, A., Kerasiotis, A., Kotsias, S., Θεοδωροπούλου, Γ., Miaoulis, G., & Ghazanfarpour, D. (2023) · Sensors · doi:10.3390/s23156931 · Cited by 22
“The most efficient path on the selected process model Predictions about the next activity and the goal state”
MethodLv, T., Hong, X., Liu, Y., Miao, K., Sun, H., Li, L., Deng, C., Jiang, C., & Pan, X. (2023) · Computer Methods and Programs in Biomedicine · doi:10.1016/j.cmpb.2023.107857 · Cited by 26
Hou, J. C., Chen, G., Li, Z., He, W., Gu, S., Knoll, A., & Jiang, C. (2023) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2023.3312647 · Cited by 21
“Reinforcement learning (RL) is a powerful method that has surpassed human performance in complex systems, such as Go [25], StarCraft [26], car racing [27], and intelligent transportation systems [28], [29].”
MethodBeyond games: a systematic review of neural Monte Carlo tree search applications
Kemmerling, M., Lütticke, D., & Schmitt, R. H. (2024) · Applied Intelligence · doi:10.1007/s10489-023-05240-w · Cited by 19
“The selection phase in AlphaZero uses a variant of the Predictor + UCT (PUCT) formula [83]: where P ( s , a ) denotes a prior probability of choosing action a in state s given by f θ [100].”
InfluentialUnconventional Inorganic‐Based Memristive Devices for Advanced Intelligent Systems
Sung, S. H., Kim, D. H., Kim, T., Kang, I., & Lee, K. J. (2019) · Advanced Materials Technologies · doi:10.1002/admt.201900080 · Cited by 20
Iterated Deep Reinforcement Learning in Games: History-Aware Training for Improved Stability
Wright, M., Wang, Y., & Wellman, M. P. (2019) · ACM Conference on Economics and Computation (EC) · doi:10.1145/3328526.3329634 · Cited by 18
“Deep RL has famously succeeded in learning superhuman strategies for extensive-form strategy games like Go, chess, and shogi [39, 40].”
BackgroundMethodBlood Bowl: A New Board Game Challenge and Competition for AI
Justesen, N., Uth, L. M., Jakobsen, C., Moore, P. D., Togelius, J., & Risi, S. (2019) · IEEE Conference on Games (CoG) · doi:10.1109/cig.2019.8848063 · Cited by 21
“It is, however, easy to modify the FFAI Gym implementation to e.g. support self-play, which is an effective learning method in two-player board games [15].”
BackgroundDeep Multi-Critic Network for accelerating Policy Learning in multi-agent environments
Hook, J., Silva, D., & KONDOZ, A. M. (2020) · Neural Networks · doi:10.1016/j.neunet.2020.04.023 · Cited by 19
Ausin, M. S., Maniktala, M., Barnes, T. M., & Chi, M. (2020) · Lecture notes in computer science · doi:10.1007/978-3-030-52237-7_38 · Cited by 25
The Impact of Artificial Intelligence on the Chess World
Iliescu, D. M. D. (2020) · JMIR Serious Games · doi:10.2196/24049 · Cited by 19
Moon, T., Lee, J. W., & Son, J. E. (2021) · Sensors · doi:10.3390/s21062187 · Cited by 21
Deliberative acting, planning and learning with hierarchical operational models
Patra, S., Mason, J., Ghallab, M., Nau, D. S., & Traverso, P. (2021) · Artificial Intelligence · doi:10.1016/j.artint.2021.103523 · Cited by 24
“For example, while the approach is adequate for a video game such as Starcraft; for board games such as chess or go, it is better to use game-tree search and learning techniques [88, 107].”
MethodIntegrating Media Selection and Media Effects Using Decision Theory
Fisher, J. T., & Hamilton, K. A. (2021) · Journal of Media Psychology Theories Methods and Applications · doi:10.1027/1864-1105/a000315 · Cited by 31
Landmark-Guided Subgoal Generation in Hierarchical Reinforcement LearningPreprint
Kim, J., Seo, Y., & Shin, J. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2110.13625 · Cited by 26
Olympiad-level formal mathematical reasoning with reinforcement learning
Hubert, T., Mehta, R., Sartran, L., Horváth, M. Z., Žužić, G., Wieser, E., Huang, A., Schrittwieser, J., Schroecker, Y., Masoom, H., et al. (2026) · Nature · doi:10.1038/s41586-025-09833-y · Cited by 19
Self-citationA novel method for intersecting machining feature segmentation via deep reinforcement learning
Zhang, H., Wang, W., Zhang, S., Zhang, Y., Zhou, J., Wang, Z., Huang, B., & Huang, R. (2023) · Advanced Engineering Informatics · doi:10.1016/j.aei.2023.102256 · Cited by 17
Lalani, S. R., Safaei, B., Monazzah, A. M. H., Taghizadeh, H., Henkel, J., & Ejlali, A. (2024) · IEEE Transactions on Green Communications and Networking · doi:10.1109/tgcn.2024.3399455 · Cited by 27
“Therefore, it has been shown that Reinforcement learning outperforms existing heuristic or rule-based approaches as well as humans [17], [18], [19].”
The Convergence of AI and animal-inspired robots for ecological conservation
Afzal, N., Rehman, M. U., Seneviratne, L. D., & Hussain, I. (2024) · Ecological Informatics · doi:10.1016/j.ecoinf.2024.102950 · Cited by 26
AI-Based Model Order Reduction Techniques: A Survey
Mallick, S., & Mittal, M. (2025) · Archives of Computational Methods in Engineering · doi:10.1007/s11831-024-10207-2 · Cited by 27
Online deep learning’s role in conquering the challenges of streaming data: a survey
Sulaiman, M. A., Farmanbar, M., Kagami, S., Belbachir, A. N., & Rong, C. (2025) · Knowledge and Information Systems · doi:10.1007/s10115-025-02351-3 · Cited by 26
Intelligent games meeting with multi-agent deep reinforcement learning: a comprehensive review
Wang, Y., Wang, Y., Tian, F., Ma, J., & Jin, Q. (2025) · Artificial Intelligence Review · doi:10.1007/s10462-025-11166-1 · Cited by 25
Chen, Y., McNeil, A. M., Park, T., Wilson, B. A., Iyer, V., Bezick, M., Choi, J.-I., Ojha, R., Mahendran, P., Singh, D. K., et al. (2025) · Nanophotonics · doi:10.1515/nanoph-2025-0049 · Cited by 25
“They recently gained popularity for their ability to beat humans at chess, Go, and other popular games [142].”
Ho, S.-H. S., Nakamura, Y., & Swain, J. E. (2021) · Frontiers in Psychology · doi:10.3389/fpsyg.2020.603385 · Cited by 25
Reward Machines for Vision-Based Robotic Manipulation
Camacho, A., Varley, J., Zeng, A., Jain, D., Işçen, A., & Kalashnikov, D. (2021) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48506.2021.9561927 · Cited by 19
“g, Atari games from pixels [1], board games [2], and robot control from pixels [3]).”
BackgroundPredicting Victories in Video Games - IEEE BigData 2021 Cup Report
Matraszek, M., Janusz, A., Świechowski, M., & Ślȩzak, D. (2021) · IEEE International Conference on Big Data (BigData) · doi:10.1109/bigdata52589.2021.9671650 · Cited by 18
“MCTS has been the staple method in the realm of combinatorial games such as Go [9], Chess [10] or Hex [11].”
MethodThe best of both worlds: Dual systems of reasoning in animals and AI
Kelly, M., & Barron, A. B. (2022) · Cognition · doi:10.1016/j.cognition.2022.105118 · Cited by 24
“AlphaGo famously beat humans in the strategy game of Go (Silver et al., 2016), and we argue that its success came from use of dual process cognitive systems (Schrittwieser et al., 2020; Silver et al., 2017; Silver et al., 2018).”
InfluentialProton gated oxide neuromorphic transistors with bionic vision enhancement and information decoding
Ren, Z. Y., Kong, Y. H., Ai, L., Xiao, H., Wang, W. S., Shi, Z. W., & Zhu, L. Q. (2022) · Journal of Materials Chemistry C · doi:10.1039/d2tc00775d · Cited by 19
Wang, C., Ding, Y., Yan, N., Ma, L., Ma, J., Lü, C., Yang, C., Su, Y., Chong, J., Jin, H., et al. (2022) · Advanced Engineering Informatics · doi:10.1016/j.aei.2022.101665 · Cited by 18
A maintenance planning framework using online and offline deep reinforcement learning
Bukhsh, Z. A., Molegraaf, H. J. A., & Jansen, N. (2023) · Neural Computing and Applications · doi:10.1007/s00521-023-08560-7 · Cited by 17
“In particular, the integration of deep learning and reinforcement learning as deep reinforcement learning (DRL) has been applied to various applications, including board games [3], video games [4], robotic control [5], and optimal routing [6].”
BackgroundMethodMulti-Agent Reinforcement Learning With Decentralized Distribution Correction
Li, K., & Jia, Q. (2024) · IEEE Transactions on Automation Science and Engineering · doi:10.1109/tase.2024.3369592 · Cited by 22
“I N THE past decade, reinforcement learning [1] has been widely applied in complex sequential decision-making tasks, e.g., playing Go [2], [3], video games [4], [5], [6], and robot control [7], [8], [9].”
Sun, C., Zang, Z., Li, J., Li, J., Xu, X., Wang, R., & Zheng, C. (2024) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v38i13.29438 · Cited by 24
Ballard, N., Farajzadehahary, K., Hamzehlou, S., Mori, U., & Asúa, J. M. (2024) · Computers & Chemical Engineering · doi:10.1016/j.compchemeng.2024.108739 · Cited by 21
Building simulation in adaptive training of machine learning models
Amini, H., Alanne, K., & Kosonen, R. (2024) · Automation in Construction · doi:10.1016/j.autcon.2024.105564 · Cited by 26
“For instance, David Silver et al. [6] presented an intelligent program playing sophisticated games (like, chess) with superhuman performance using a deep neural network (DNN) and reinforcement learning (RL) by self-play.”
BackgroundTeaching Machines to Recognize Neurodynamic Correlates of Team and Team Member Uncertainty
Stevens, R. H., & Galloway, T. L. (2019) · Journal of Cognitive Engineering and Decision Making · doi:10.1177/1555343419874569 · Cited by 17
“…functions to efficiently perform tasks that humans can describe like image processing, facial recognition, credit card verification, vehicle navigation, beating chess champions, and most recently, teaching themselves to master board games through self-play (Hsu, 2002; Silver et al., 2018).”
BackgroundAgent Coordination in Air Combat Simulation using Multi-Agent Deep Reinforcement Learning
Källström, J., & Heintz, F. (2020) · Conference proceedings/Conference proceedings - IEEE International Conference on Systems, Man, and Cybernetics · doi:10.1109/smc42975.2020.9283492 · Cited by 21
“By combining reinforcement learning with deep learning it has become possible to achieve impressive results in complicated control tasks [4]–[6], classic board games [7]–[9], and challenging real-time, multiplayer computer games [10], [11].”
BackgroundScheinker, A. (2021) · Information · doi:10.3390/info12040161 · Cited by 20
TCP-NeuRoc: Neural Adaptive TCP Congestion Control With Online Changepoint Detection
Li, W., Gao, S., Li, X., Xu, Y., & Lu, S. (2021) · IEEE Journal on Selected Areas in Communications · doi:10.1109/jsac.2021.3087247 · Cited by 21
Basic Study for Transfer Learning for Autonomous Driving in Car Race of Model Car
Chiba, S., & Sasaoka, H. (2021) · International Conference on Business and Industrial Research (ICBIR) · doi:10.1109/icbir52339.2021.9465856 · Cited by 21
What will drive global economic growth in the digital age?
Growiec, J. (2022) · Studies in Nonlinear Dynamics and Econometrics · doi:10.1515/snde-2021-0079 · Cited by 21
Hwang, H.-S., Lee, M., & Seok, J. (2022) · Applied Soft Computing · doi:10.1016/j.asoc.2022.109386 · Cited by 19
Artificial Intelligence in the 21st Century
奥宮, 正. (2023) · Research on Intelligent Manufacturing and Assembly · doi:10.25082/rima.2023.01.002 · Cited by 21
EgoMap: Projective Mapping and Structured Egocentric Memory for Deep RL
Beeching, E., Dibangoye, J., Simonin, O., & Wolf, C. (2021) · Lecture notes in computer science · doi:10.1007/978-3-030-67661-2_31 · Cited by 27
Drori, I., Krishnamurthy, Y., Lourenço, R., Rampin, R., Cho, K., Silva, C. T., & Freire, J. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1905.10345 · Cited by 36
“AlphaZero (Silver et al., 2018) reduced this search by three orders of magnitude using a general reinforcement learning algorithm.”
MethodLearning to Play No-Press Diplomacy with Best Response Policy IterationPreprint
Anthony, T., Eccles, T., Tacchetti, A., Kramár, J., Gemp, I., Hudson, T. C., Porcel, N., Lanctot, M., Pérolat, J., Everett, R., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2006.04635 · Cited by 57
“We adopt a policy iteration (PI) based approach, motivated by successes using PI for perfect information, sequential move, 2-player zero-sum board games [2, 109].”
BackgroundMethodInfluentialSelf-citationReinforcement Learning on Job Shop Scheduling Problems Using Graph Networks.Preprint
Hameed, M. S. A., & Schwung, A. (2020) · arXiv (Cornell University) · Cited by 18
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
Asymmetric self-play for automatic goal discovery in robotic manipulation
OpenAI, O., Plappert, M., Sampedro, R., Xu, T., Akkaya, I., Kosaraju, V., Welinder, P., D'Sa, R., Petron, A., Pinto, H. P. D. O., et al. (2021) · arXiv · doi:10.48550/arxiv.2101.04882 · Cited by 18
Tabas, S. S., & Samadi, V. S. (2024) · Journal of Water Resources Planning and Management · doi:10.1061/jwrmd5.wreng-6089 · Cited by 17
Shalu, Verma, N., Dev, K., Bhardwaj, A. B., & Kumar, K. (2025) · Annals of Neurosciences · doi:10.1177/09727531251359872 · Cited by 42
Ludii as a Competition Platform
Stephenson, M., Piette, É., Soemers, D. J. N. J., & Browne, C. (2019) · IEEE Conference on Games (CoG) · doi:10.1109/cig.2019.8848084 · Cited by 18
“This includes many board or card games such as Chess [3], Go [4], Shogi [5], Checkers [6], Backgammon [7], Hanabi [8] and Poker [9], as well as popular video games such as Super Mario Bros. [10], Starcraft [11], Angry Birds [12], Ms. PacMan [13], Doom [14] and Unreal Tournament [15].”
BackgroundJameel, F., Sharma, N., Khan, M. A., Khan, I., Alam, M. M., Mastorakis, G. N., & Mavromoustakis, C. X. (2020) · Internet of things · doi:10.1007/978-3-030-44907-0_8 · Cited by 17
RePReL: Integrating Relational Planning and Reinforcement Learning for Effective Abstraction
Kokel, H., Manoharan, A., Natarajan, S., Ravindran, B., & Tadepalli, P. V. (2021) · Proceedings of the International Conference on Automated Planning and Scheduling · doi:10.1609/icaps.v31i1.16001 · Cited by 18
“Here explicit dynamic models of actions are learned via exploration and used either offline to compute approximately optimal policies (Brafman and Ten-nenholtz 2002; Guestrin, Patrascu, and Schuurmans 2002) or online in look-ahead search (Silver, Hubert et al. 2018).”
MethodLBGP: Learning Based Goal Planning for Autonomous Following in Front
Nikdel, P., Vaughan, R., & Chen, M. (2021) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48506.2021.9560914 · Cited by 19
Yin, L., & Wu, Y. (2021) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2021.104508 · Cited by 16
Unlocking the Flexibility of District Heating Pipeline Energy Storage with Reinforcement Learning
Stepanovic, K., Wu, J., Everhardt, R., et al. (2022) · Energies · doi:10.3390/en15093290 · Cited by 17
“A common approach in the literature is to incorporate objective functions and safety constraints into the reward function [34].”
BackgroundResonance for Analog Recurrent Neural Network
Qu, Y., Zhou, M., Khoram, E., Yu, N., & Yu, Z. (2022) · ACS Photonics · doi:10.1021/acsphotonics.1c02016 · Cited by 18
Wang, C., Han, C., Guo, T., & Ding, M. (2022) · Applied Intelligence · doi:10.1007/s10489-022-03453-z · Cited by 20
“Another standout work was the success of AlphaGo [18], AlphaZero [19] and Muzero [20] which are a series of works on mastering board game and video games.”
BackgroundDeep Learning for Voltammetric Sensing in a Living Animal Brain
Xue, Y., Ji, W., Jiang, Y. Y., Yu, P., & Mao, L. (2021) · Angewandte Chemie · doi:10.1002/ange.202109170 · Cited by 17
Metamorphic relations via relaxations: an approach to obtain oracles for action-policy testing
Enişer, H. F., Gros, T. P., Wüstholz, V., Hoffmann, J., & Christakis, M. (2022) · ACM SIGSOFT International Symposium on Software Testing and Analysis (ISSTA) · doi:10.1145/3533767.3534392 · Cited by 20
“Action policies represented by neural networks are highly successful in complex, sequential decision making problems, specifically in games [32, 38, 39] and increasingly in Artificial Intelligence (AI) Planning [16, 19, 24, 25, 43].”
BackgroundTackling Real-World Autonomous Driving using Deep Reinforcement Learning
Maramotti, P., Capasso, A. P., Bacchiani, G., & Broggi, A. (2022) · IEEE Intelligent Vehicles Symposium · doi:10.1109/iv51971.2022.9827302 · Cited by 21
Techniques and Paradigms in Modern Game AI Systems
Lu, Y., & Li, W. (2022) · Algorithms · doi:10.3390/a15080282 · Cited by 21
“The same framework is used to train AlphaZero [4], achieving superhuman performance in Go, Chess, and Shogi.”
BackgroundMethodPylorof, D., & Garcia, H. E. (2022) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2022.105454 · Cited by 20
Reinforcement Learning for Multiaircraft Autonomous Air Combat in Multisensor UCAV Platform
Kong, W., Zhou, D., Du, Y., Zhou, Y., & Zhao, Y. (2022) · IEEE Sensors Journal · doi:10.1109/jsen.2022.3220324 · Cited by 20
“The competitive SP method [39] is a method that learning policy that plays against oneself without any direct supervision yields excellent performance in challenging tasks such as Go, Poker, Simulated Sumo, Dota 2, StarCraft II, and more.”
MethodAdvanced Reinforcement Learning and Its Connections with Brain Neuroscience
Fan, C., Li, Y., Zhang, J., Zhen, Z., & Wu, X. (2023) · Research · doi:10.34133/research.0064 · Cited by 26
Wong, Y. J., Tham, M., Kwan, B.-H., & Owada, Y. (2023) · Sensors · doi:10.3390/s23052494 · Cited by 23
Hafez, M. B., Immisch, T., Weber, T., & Wermter, S. (2023) · Frontiers in Neurorobotics · doi:10.3389/fnbot.2023.1127642 · Cited by 22
“This includes solving the game of Go (Silver et al., 2016) and other challenging games (Silver et al., 2018; Schrittwieser et al., 2020), as well as achieving breakthroughs in robotic manipulation (Levine et al., 2016), locomotion (Haarnoja et al., 2019), and navigation (Zhu et al., 2017).”
BackgroundMethodInfluentialYu, G., Wu, C., Liu, Y., Du, Z., Zhao, N., Jiao, Y., Xu, Z., & Shi, W. (2023) · Science China Earth Sciences · doi:10.1007/s11430-022-1144-7 · Cited by 20
A test of memory for stimulus sequences in great apes
Lind, J., Vinken, V., Jonsson, M., Ghirlanda, S., & Enquist, M. (2023) · PLoS ONE · doi:10.1371/journal.pone.0290546 · Cited by 18
Kim, H.-S., Park, H., & Cho, W.-J. (2023) · Biomimetics · doi:10.3390/biomimetics8070532 · Cited by 19
“Recent advances in artificial intelligence (AI) technology have revolutionized various aspects of our daily lives, such as visual and language recognition as well as automatic driving [1–3].”
BackgroundLin, K., Li, Y., Liu, Q., Li, D., Shi, X., & Chen, S. (2024) · Information Sciences · doi:10.1016/j.ins.2024.120261 · Cited by 20
A sequence bottleneck for animal intelligence and language?
Lind, J., & Jon-And, A. (2024) · Trends in Cognitive Sciences · doi:10.1016/j.tics.2024.10.009 · Cited by 19
“Although models can incorporate faithful representations of stimulus sequences, such as in powerful arti fi cial intelligence (AI) systems [8], computational approaches in both animal and machine learning have shown that complex behavioral sequences can be acquired without requiring faithful…”
Bayes-Adaptive Monte-Carlo Planning and Learning for Goal-Oriented Dialogues
Jang, Y., Lee, J., & Kim, K.-E. (2020) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v34i05.6308 · Cited by 17
Jameel, F., Javaid, U., Sikdar, B., Khan, I., Mastorakis, G. N., & Mavromoustakis, C. X. (2020) · Internet of things · doi:10.1007/978-3-030-44907-0_12 · Cited by 19
Artificial Intelligence – Challenges and Chances for Europe
Straus, J. (2020) · European Review · doi:10.1017/s1062798720001106 · Cited by 18
Kurita, M., & Hoki, K. (2020) · IEEE Transactions on Games · doi:10.1109/tg.2020.3036471 · Cited by 21
“, Backgammon [7], Checkers [8], Chess [9], Shogi [10], [11], and Go [12].”
BackgroundMastrogiuseppe, C., & Moreno‐Bote, R. (2022) · Scientific Reports · doi:10.1038/s41598-022-13862-2 · Cited by 23
Sample Efficient Reinforcement Learning via Model-Ensemble Exploration and Exploitation
Yao, Y., Li, X., An, Z., Zhang, W., & Luo, D. (2021) · Proceedings - IEEE International Conference on Robotics and Automation/Proceedings · doi:10.1109/icra48506.2021.9561842 · Cited by 34
“Reinforcement learning (RL) has achieved many remarkable results in recent years, including outperforming human performances on video games [1]–[3], mastering the game of GO [4], playing card games [5], and even learning manipulation skills for robotics [6], [7].”
BackgroundDouZero+: Improving DouDizhu AI by Opponent Modeling and Coach-guided Learning
Zhao, Y., Zhao, J., Hu, X., Zhou, W., & Li, H. (2022) · IEEE Conference on Games (CoG) · doi:10.1109/cog51982.2022.9893710 · Cited by 18
Obtaining Human Experience for Intelligent Dredger Control: A Reinforcement Learning Approach
Wei, C., Ni, F., & Chen, X. (2019) · Applied Sciences · doi:10.3390/app9091769 · Cited by 18
“The virtual environment can ensure that the learning agent can explore the entire state space and the action space as in playing games in [15,16].”
BackgroundPlanning With Uncertain Specifications (PUnS)
Shah, A., Li, S., & Shah, J. (2020) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2020.2977217 · Cited by 22
“Prior research into reinforcement learning has indicated great promise in sequential decision-making tasks, with breakthroughs in handling large-dimensional state spaces such as Atari games [11], continuous action spaces ([12], [13]), sparse rewards ([14], [15]), and all of these challenges in combination [16].”
BackgroundAssessing Game Balance with AlphaZero: Exploring Alternative Rule Sets in ChessPreprint
Tomašev, N., Paquet, U., Hassabis, D., & Kramnik, V. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2009.04374 · Cited by 46
“AlphaZero is an adaptive learning system that improves through many rounds of self-play (Silver et al., 2018).”
BackgroundMethodInfluentialSelf-citationBeBold: Exploration Beyond the Boundary of Explored RegionsPreprint
Zhang, T., Xu, H., Wang, X., Wu, Y., Keutzer, K., Gonzalez, J. E., & Tian, Y. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2012.08621 · Cited by 49
“Deep reinforcement learning (RL) has experienced significant progress over the last several years, with impressive performance in games like Atari (Mnih et al., 2015; Badia et al., 2020a), Star-Craft (Vinyals et al., 2019) and Chess (Silver et al., 2016; 2017; 2018).”
BackgroundAlgorithmic and Human CollusionPreprint
Werner, T. (2021) · SSRN Electronic Journal · doi:10.2139/ssrn.3960738 · Cited by 17
Metadata
- Authors
- Silver, D., Hubert, T., Schrittwieser, J., Antonoglou, I., Lai, M., Guez, A., Lanctot, M., Sifre, L., Kumaran, D., Graepel, T., Lillicrap, T., Simonyan, K., & Hassabis, D.
- Affiliations
- Google DeepMind (United Kingdom) (GB)
- Corresponding author
- Silver, David
- Hubert, Thomas
- Schrittwieser, Julian
- Hassabis, Demis
- ORCID
- Venue
- Science
- Published
- 2018-12-07
- Volume
- 362
- Issue
- 6419
- Pages
- 1140-1144
- ISSN
- 0036-8075 (ISSN-L)
- 1095-9203
- OpenAlex
- W2902907165
- Cited by
- 3,562
- References
- 30
- Citations by year
- 2026: 257 · 2025: 427 · 2024: 512 · 2023: 514 · 2022: 569 · 2021: 546 · 2020: 462 · 2019: 256 · 2018: 13
- FWCI
- 179.89
- Type
- journal article
- Language
- en
- Open access
- bronze
- doi.org
- Locations
- Science (publishedVersion)
- PubMed (publishedVersion)
- UCL Discovery (University College London) (submittedVersion)
- Funders
- DeepMind
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Artificial Intelligence in Games
- Physical Sciences › Computer Science › Artificial Intelligence › Reinforcement Learning in Robotics
- Physical Sciences › Computer Science › Artificial Intelligence › Artificial Intelligence Applications
- Keywords
- AlphaZero, chess, shogi, Go, evaluation function learning, game tree search
- MeSH terms
- Algorithms, Artificial Intelligence, Humans, Reinforcement, Psychology, Software, Video Games
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.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
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
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.