A general reinforcement learning algorithm that masters chess, shogi, and Go through self-play
Open accessThe 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.
Abstract
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 3,501–3,600 of 3,633
Learn-to-Probe: Achieving Signal Distinguishability in Learning-based Congestion Control
Tian, H., Li, W., Zhang, J., et al. (2026) · European Conference on Computer Systems · doi:10.1145/3767295.3769317
TREND: Task-Oriented World Models for Visual Robotic Manipulation
Zheng, Y., Lu, T., & Cai, Y. (2026) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2026.3688085
“R EINFORCEMENT Learning (RL) addresses decision-making tasks such as board games [1] [2], autonomous driving [3] [4], robotic manipulation [5] [6] and fine-tuning of large language models [7] [8] by learning from past interactions, but typically requires extensive environment interactions.”
Kelters, B. (2026) · AI & Society · doi:10.1007/s00146-026-03055-0 · Cited by 1
“Some such machines plausibly already exist, seemingly matching or exceeding human intellectual performance in certain areas (see, for examples, Silver et al. 2018, p. 1140; Martínez 2024 and Tomlinson et al. 2024).”
Yao, X., Hui, K., & Xu, Z. (2026) · Lecture notes in electrical engineering · doi:10.1007/978-981-95-7656-2_51
Mapping AAV capsid sequences to functions through function-guided in silico evolution
Guo, B., Zheng, H., Lin, X., Jiang, H., Mo, A., Wei, H., Zhang, R., Yuan, Y., Wu, Y., Li, H., et al. (2026) · Cell Press Blue · doi:10.1016/j.cpblue.2026.100016 · Cited by 3
Efficient learning for active wrist compensation strategy of ledge-climbing robot
Aji, A. A. K., & Lin, C.-Y. (2026) · Applied Intelligence · doi:10.1007/s10489-026-07245-7
Meta-ethics and AI: exploring the novel meta-ethical questions in the era of AI
Lu, S. (2026) · AI and Ethics · doi:10.1007/s43681-026-01134-y
“By contrast, when AlphaGo and its successors began to defeat top human players in Go—first with human-guided learning [103 ] and later via self-play that produced novel strategies [104 ]—the pattern of explanation shifted.”
Ghojogh, B., & Ghodsi, A. (2026) · Springer eBooks · doi:10.1007/978-3-032-10738-1_18
QFlowNet: Fast, Diverse, and Efficient Unitary Synthesis with Generative Flow Networks
Koo, I., Cha, H., & Lee, J. (2026) · International Conference on Quantum Communications, Networking, and Computing (QCNC) · doi:10.1109/qcnc69040.2026.00098 · Cited by 2
“Compared to deep RL approaches like Gum-bel AlphaZero [25], [27], our QFlowNet method is more computationally feasible.”
ProdNet: A Lightweight Network for Fast Discovery of Matrix Multiplication Algorithms
Lakas, B. O., Berdjouh, C., Bounane, K., Kherfi, M. L., Aiadi, O., & Belhouari, S. B. (2026) · Communications in computer and information science · doi:10.1007/978-3-032-21585-7_7
AGYLIS: A Generative Human Simulation Framework via Human–Avatar Behavioral Comparison in VR
Tanikawa, T., Ando, M., & Nakagawa, S. (2026) · Augmented Humans International Conference (AHs) · doi:10.1145/3795011.3797395
“Furthermore, goal-based AI has been proposed, which performs goal-based planning and search as a method for more advanced decision-making[2–4].”
PPTopoGym: Towards an RL Environment for Topology Actions on Power Grids
Köhler, D., Hassouna, M., Degtyar, D., Krauß, J., Brendlinger, K., & Scholz, C. (2026) · Communications in computer and information science · doi:10.1007/978-3-032-19102-1_6
Adaptable Hindsight Experience Replay for Search-Based Learning
Vazaios, A., Brugger, J., Derstroff, C., Kersting, K., & Mezini, M. (2026) · Communications in computer and information science · doi:10.1007/978-3-032-19105-2_2
Self-learning Chess Bot Using Monte Carlo Tree Search Technique
Srivani, M., Sheshang, R., & Pandiyan, C. A. N. (2026) · Lecture notes in networks and systems · doi:10.1007/978-981-95-3492-0_13
Reinforcement learning based data assimilation for unknown state model
Wang, Z., Jiang, L., & He, X. (2026) · Journal of Computational and Applied Mathematics · doi:10.1016/j.cam.2026.117777 · Cited by 1
Speed and quality of complex strategic decisions
Sunde, U., Zēgners, D., & Strittmatter, A. (2026) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.2531472123 · Cited by 1
“Recently, more general machine learning techniques in the form of neural networks have been embodied in chess engines such as Google’s nonpublic engine AlphaZero (53), the open-source engine LeelaChessZero, or the Maia Chess project, which trains a neural network designed to capture a human-like style of playing.”
ITLMS: an intelligent traffic light management system using edge computing and deep learning
Noor, A., Ziad, A., Alharbi, B., Alharbi, G., & Noor, T. H. (2026) · PeerJ Computer Science · doi:10.7717/peerj-cs.3858
Probing Emergent World Representations in Go Life-and-Death Problems
Yang, Z., Meng, Z., & Wen, Z. (2026) · AI · doi:10.3390/ai7050170
Autonomous Maneuver Decision‐Making for Unmanned Aerial Vehicles via Transfer Reinforcement Learning
Huang, Y., Long, Y., Cheng, Z., Shen, A., & Liu, M. H. (2026) · International Journal of Robust and Nonlinear Control · doi:10.1002/rnc.70586
Moyá, A. (2026) · Springer eBooks · doi:10.1007/978-3-032-25682-9_6
Agent-based ecosystem modeling with deep reinforcement learning
Strannegård, C., Palak, M., Engsner, N., Stocco, A., Antonelli, A., & Silvestro, D. (2026) · Ecological Informatics · doi:10.1016/j.ecoinf.2026.103819 · Cited by 2
Move Quality Beyond Arena: Pure Distillation for Small Go NetworksPreprint
Kojima, T. (2026) · Research Square · doi:10.21203/rs.3.rs-9749016/v1
Zhang, J., Chen, L., & Benavides-Riveros, C. L. (2026) · Journal of Chemical Theory and Computation · doi:10.1021/acs.jctc.6c00988
Mastering Othello with Genetic Algorithm and Reinforcement Learning
Okashita, Y., & Sudo, Y. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-23657-9_9
Relevance-Zone Reduction in Game Solving
Lin, C. Y., Wei, T.-H., Wang, C., et al. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-23657-9_1
Evolutionary Tree Search for Turn-Based Strategy Games
Neller, T., Huang, L., & Cho, J.-H. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-23657-9_12
Tari, H. (2026) · Journal of vibration and acoustics · doi:10.1115/1.4072014
Wang, X., Luo, B., Hu, T., Cui, Y., & Feng, Y. (2026) · Systems Science & Control Engineering · doi:10.1080/21642583.2026.2672169 · Cited by 5
Overcoming Challenges of Realism in Competitive Space-Based Reinforcement Learning with AstroCraft
Anderson, L. J., Erwin, R. S., Hansen, M., Outkin, V., Qureshi, R. S., Aborizk, A., Kulkarni, R., Dennery, E., Senapathy, S., Makkapati, S., et al. (2026) · IEEE Aerospace Conference (AERO) · doi:10.1109/aero66936.2026.11519822 · Cited by 1
“Some works have trained using self-play from scratch to prevent this initial bias, such as TD-Gammon [9] on backgammon, AlphaGo Zero [10] on Go, and AlphaZero [11] on Go, Chess, and Shogi.”
Rabczuk, T., Anitescu, C., Goswami, S., et al. (2026) · Studies in systems, decision and control · doi:10.1007/978-3-032-20307-6_1
Discerning Quantity: Numerosity in Two Embodied Machine Learning Agents
Donnelly, N., & Keedwell, E. C. (2026) · Behavioral Sciences · doi:10.3390/bs16050813
Overview of Intelligent Game: From Game AI to Military Combat Simulation
Ma, W., Xiao, Y., Wang, L., et al. (2026) · International Journal of Crowd Science · doi:10.26599/ijcs.2024.9100027
PokerOWL: A Multi-Agent Poker Environment for Benchmarking Open-World Learning
Chiu, M., Nananukul, N., & Kejriwal, M. (2026) · Applied Sciences · doi:10.3390/app16115458
Mou, Z., Tian, Z., Jin, J., Yu, H., & Huang, Y. (2026) · Biomolecules · doi:10.3390/biom16060830 · Cited by 1
“Even before the concept of LLMs emerges, agents are constructed using rule matching [114], reinforcement learning [115], and other methods are already applied in academia and industry for auxiliary decision-making in complex environments.”
A General Upper Bound for the Runtime of a Coevolutionary Algorithm on Impartial Combinatorial Games
Benford, A., & Lehre, P. K. (2026) · Algorithmica · doi:10.1007/s00453-026-01397-1 · Cited by 1
“…games are a long-standing focus in the development of artificial intelligence, from Donald Michie’s seminal use of reinforcement learning on Tic-Tac-Toe [46], to Deep Blue’s famous matches against then-world Chess champion Garry Kas-parov [10], to the recent groundbreaking results of DeepMind [61].”
Lee, C. T., & Shen, Y. (2026) · Psychology and Marketing · doi:10.1002/mar.70184
“Technological advancements have enabled AI to function as a formidable adversary (Silver et al. 2018), making competitive interactions with AI feasible and salient for consumers.”
InfluentialHybrid MCTS++: An Adaptive Optimizer for Enhanced Game-Playing Performance
Abuhashish, F., Atoum, M., & Ismail, I. H. (2026) · doi:10.1109/icciaa68481.2026.11543971
“The Alp-haZero algorithm uses a softmax policy to compute priors and selects actions by sampling from visit counts [2].”
InfluentialA mechanism-aided transformer may transform in-flight aviation safety
Song, J., Hou, F., Wang, X. Z., & Yue, X. (2026) · PNAS Nexus · doi:10.1093/pnasnexus/pgag154
DRL-based Flexible Control That Can Adapt to Changes in Aircraft Configuration
Sarıgül, F. A., & Bayezit, İ. (2026) · International Conference on Control, Automation and Robotics (ICCAR) · doi:10.1109/iccar69571.2026.11549622
“Such a flight control approach is offered through a DRL-based control strategy that utilize the generalization ability Artificial Intelligence (AI) [1].”
The 2025 Computer Olympiad: Athénan Wins 9 Gold Medals
Cohen-Solal, Q., & Cazenave, T. (2026) · ICGA Journal · doi:10.1177/13896911261455180
“Unlike AlphaZero-like algorithms (Silver et al., 2018), Athénan is based on the Descent framework (Cohen-Solal, 2020).”
Elkarnighi, K., Alaoui, A. O. E., Sarhani, M., et al. (2026) · Journal of the Knowledge Economy · doi:10.1007/s13132-026-03315-w
“RL has been successfully applied in fields such as robotics and games (Silver et al., 2018).”
Explainability and Interpretability in Machine Learning and Artificial Intelligence
Rowat, C. (2026) · Oxford University Press eBooks · doi:10.1093/9780197851418.003.0994
Wang, T. R., Yang, Y., & Zhou, J. (2026) · International Journal of Human-Computer Interaction · doi:10.1080/10447318.2026.2686773 · Cited by 2
Sensory horizons, reality monitoring, and the higher-order nature of consciousness: A critical look
Axel Cleeremans, B., & Dołęga, K. (2026) · Behavioral and Brain Sciences · doi:10.1017/s0140525x25103415
PhyZero: Physics-Grounded Zero-Shot Driving Agent
Sun, J., Liu, H., & Zhang, T. A. (2026) · Chinese Control and Decision Conference (CCDC) · doi:10.1109/ccdc69976.2026.11559983
“The selection is guided by the Predictor Upper Confidence Bound (PUCT) algorithm [11], which balances exploitation (high value) and exploration (low visit count), incorporated with a prior policy: (2) where Q ( s, a ) is the estimated action value, N ( s, a ) is the visit count, and P ( s, a )…”
A Tool for Visualizing the Physical Elements of Communities
Wagner, S., Diamond, S., Kossowsky, D., Wang, C. S., Klimczak, J., & Roorda, M. J. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-30044-7_24
Kissina, M., Dyussembayeva, A., Saimanova, Z., et al. (2026) · Trudy Universiteta · doi:10.52209/1609-1825_2026_2_464
Wang, K., Ma, D., Yang, M., et al. (2026) · Aerospace Science and Technology · doi:10.1016/j.ast.2026.113071
Stack-Aware Enemy AI in a Roguelike Deck-Builder
Taneja, T., & Robertson, J. (2026) · doi:10.1145/3807895.3807928
“…in adversarial AI gameplay, which can achieve superhuman performance in complex games using deep reinforcement learning [Mnih et al. 2015] and Monte Carlo Tree Search [Browne et al. 2012] as seen in AlphaZero [Silver et al. 2018], creating challenging AI opponents for games is a difficult problem.”
Failure-Based Testing for Deep Reinforcement Learning Agents
Lin, W., Meng, J., & Zheng, Z. (2026) · Proceedings of the ACM on software engineering. · doi:10.1145/3808185 · Cited by 2
Gamma Knife treatment planning using knowledge‐based reinforcement learning
Huynh, C., Ahlgren, B., Zhang, B., et al. (2026) · Medical Physics · doi:10.1002/mp.70546
Dharangaonkar, S., Kataria, A., & Mundhe, S. (2026) · Lecture notes in networks and systems · doi:10.1007/978-3-032-08600-6_20
Neural-MCTS Test Prioritization for Smart Contract Mutation Testing
Barboni, M., Lampa, F., Morichetta, A., & Polini, A. (2026) · Distributed Ledger Technologies Research and Practice · doi:10.1145/3828728
“This approach, introduced in the context of AlphaZero [36], replaces random rollouts with learned estimates obtained from two neural networks: a policy network 𝑝 𝜃 (𝑠) , which outputs a probability distribution over actions in state 𝑠 , and a value network 𝑣 𝜃 (𝑠) , which estimates the…”
Influential138139Chapter 6 Computational Advances in Reinforcement Learning
Lizy, R. F. S., Susan, V. S., Sivarethinamohan, R., et al. (2026) · De Gruyter eBooks · doi:10.1515/9783111633909-006
Long, H., & Zhao, G. (2026) · International Conference on Computer Supported Cooperative Work in Design (CSCWD) · doi:10.1109/cscwd68734.2026.11581591
A survey of modern neural network based methods for the graph coloring problem
Tan, J. Y., Sim, K. A., & Chua, H. N. (2026) · Neural Networks · doi:10.1016/j.neunet.2026.109320 · Cited by 1
Low Compute End-to-End Robot Navigation from Depth Images Using Deep Reinforcement Learning
Hussain, S., Wahid, A., & Anayatullah, M. (2026) · International Journal of Innovations in Science and Technology · doi:10.33411/ijist/1816
Adhikari, T. N., Sayers, W., & Zhang, S. (2026) · Biomimetics · doi:10.3390/biomimetics11070476
“Deep RL [22,31] extends these principles to high-dimensional decision spaces, with large-scale demonstrations in complex sequential domains [32].”
Generation and Evaluation of Exploration Terms for Monte Carlo Tree Search
Cazenave, T. (2026) · ICGA Journal · doi:10.1177/13896911261463788
“This is the search algorithm used in AlphaGo (Silver et al., 2016), AlphaGo Zero (Silver et al., 2017), AlphaZero (Silver et al., 2018), and MuZero (Schrittwieser et al., 2020).”
Evolving Macro-Actions for Monte Carlo Tree Search in Real-Time Domains
Gourley, A., Dann, M., Li, X., et al. (2026) · Genetic and Evolutionary Computation Conference (GECCO) · doi:10.1145/3795095.3805064
“Forward search techniques for decision-making have achieved notable success in a range of complex domains, most famously in games such as Chess and Go [20], where Monte Carlo Tree Search (MCTS) has achieved superhuman performance.”
Enhancing Entertainment Through Reinforcement Learning: Applications in Gaming
Mustapha, A. A., Yoosuf, M. S., Selvarajan, S., Khadidos, A., & Khadidos, A. O. (2026) · International Journal of Computer Games Technology · doi:10.1155/ijcg/3031640
Tedeschi, G., Mussi, M., Metelli, A. M., & Restelli, M. (2026) · Machine Learning · doi:10.1007/s10994-026-07085-z
“…& Barto, 2018 ) has achieved remarkable success across a wide spectrum of complex real-world control tasks, such as robotic manipulation (Peters & Schaal, 2006), autonomous driving (Likmeta et al., 2020), game playing (Silver et al., 2018 ), and financial portfolio management (Jiang & Liang, 2017).”
Making Nuclear Energy More Competitive for Future Energy Markets with Thermal Hydraulics Research
Yoon, C., Baek, J. M., Hwang, J. H., et al. (2026) · Nuclear Technology · doi:10.1080/00295450.2026.2671549
“Unlike games such as chess, Go, or Atari, which have clearly defined winning conditions [32], severe nuclear accident research does not have explicitly defined “winning conditions.””
Efficient Decentralized Multi-UAV Wildfire Monitoring via MCTS-Distilled Diffusion Policies
Wang, L., & Bronz, M. (2026) · International Conference on Unmanned Aircraft Systems (ICUAS) · doi:10.1109/icuas69441.2026.11598691
“In works like AlphaZero [16], MCTS is combined with deep learning to guide decision-making, achieving superhuman performance in complex games by using self-play and policy/value networks.”
Nasuta, A., Zhou, H. A., Abdelrazeq, A., & Schmitt, R. H. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-25032-2_4
Eff-Jiu: Optimizing MCTS for Tibetan Jiu Chess with Prior Knowledge and Efficient Search Techniques
Gu, F., Xu, H., Wang, Y., Liu, S., & Li, X. (2026) · Lecture notes in computer science · doi:10.1007/978-981-92-3394-6_25
Prakash, D., Agrawal, A., Khalid, S., et al. (2026) · International Journal of Intelligent Robotics and Applications · doi:10.1007/s41315-026-00554-7
Transferring FX Features and WDLP Heads to Accelerate AlphaZero for Gomoku
Hossain, M. R., Liu, X., & Li, C. (2026) · Lecture notes in computer science · doi:10.1007/978-981-92-3381-6_50
A dual GCN–MCTS framework for yield-aware scheduling in multistage manufacturing
Lee, S. M., & Lee, D.-H. (2026) · Computers & Operations Research · doi:10.1016/j.cor.2026.107606
Zhao, J., Yu, Y., Wu, G., Chen, L., & Zhu, Q. (2026) · Smart Agricultural Technology · doi:10.1016/j.atech.2026.102418
57Chapter 3 Key Technologies Driving the Agentic AI Paradigm
Rakhi, S., Singhal, N., Saxena, S., & Sundaravadivazhagan, B. (2026) · De Gruyter eBooks · doi:10.1515/9783112223062-003
Huang, B., Li, J., Li, Z., Jiang, H., Yang, M., Li, X., & Niu, X. (2026) · Information · doi:10.3390/info17070708 · Cited by 1
“…old ← θ // Save the current parameters 16: Total loss L ← 0 17: for each sample ( o , a 1 . . . a T , r 1 . . . r T , π 1 . . . π T , v 1 . . . v T ) in the batch do 18: The MCTS in this study follows the standard iterative process of “selection, expansion, simulation, and backpropagation” [12,18].”
Characterizing robustness of strategies to novelty in zero-sum open worlds
Kejriwal, M., Thomas, S., & Li, H. (2026) · Academia AI and Applications · doi:10.20935/acadai8424
Playing in the Dark: Invisible Chess as a Laboratory for Strategic AI
Ciancarini, P. (2026) · AI & Innovation · doi:10.1002/aiv2.70012
Adnyana, K., & Schwung, A. (2026) · Machine Learning with Applications · doi:10.1016/j.mlwa.2026.100960
Optimized Kernels in MobileNet Architectures for Computer Chess
Mathiot, R., Goudet, O., & Cazenave, T. (2026) · ICGA Journal · doi:10.1177/13896911261467835
“Deep reinforcement learning frameworks, embodied by AlphaZero, combine policy and value networks with Monte Carlo tree search (MCTS) to achieve superhuman performance in chess and other games (Silver et al., 2018, 2017).”
Scalable decentralized controllers for cohesive navigation using graph reinforcement learning
Omotuyi, O., & Kumar, M. (2026) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2026.105667
From Gameplay Traces to Game Mechanics: Causal Induction with Large Language Models
Jiwatode, M., Dockhorn, A., & Rosenhahn, B. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-31927-2_1 · Cited by 2
Rawat, P., & Rust, J. P. (2026) · Oxford Research Encyclopedia of Economics and Finance · doi:10.1093/9780197851418.003.0990
Reinforcement learning in dynamic pricing: a survey
Kastius, A., Hagedorn, C., & Schlößer, R. (2026) · Journal of Revenue and Pricing Management · doi:10.1057/s41272-026-00619-z
Learning-Support Method for Professional Shogi Players Using Differences in Intuition with Others
Isaka, T., & Toshima, I. (2026) · Communications in computer and information science · doi:10.1007/978-3-032-21997-8_16
Kojima, T. (2026) · Research Square · doi:10.21203/rs.3.rs-10631240/v1
Critic as a Lyapunov Function: A Case Study on an Autonomous Underwater Vehicle
Bolychev, A., Malaniya, G., & Osinenko, P. (2026) · International Siberian workshops and tutorials on electron devices and materials · doi:10.1109/edm69524.2026.11632249
“Reinforcement learning (RL) has demonstrated remarkable success in complex control problems, including board games [1], real-time strategy games [2], and robotic manipulation [3].”
Designing biochemical circuits with tree searchPreprint
Bhamidipati, P. S., & Thomson, M. W. (2026) · eLife · doi:10.7554/elife.106497.2
Learning Tactical Tank Behaviour: From Reward Hacking to Robust Alignment
Krieg, M., & Hlavacs, H. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-33182-3_8
Niketan, N. (2026) · Lecture notes in networks and systems · doi:10.1007/978-3-032-32726-0_39
Multi-modal classification based on Monte Carlo tree search with Pareto optimality guarantee
Liu, G., Qian, Y., Qi, X., et al. (2026) · Pattern Recognition · doi:10.1016/j.patcog.2026.114653
Shivkumar, H., Bhatlapenumarthi, P., Gupta, S., Srivastava, T., Narayanan, S. J., & Perumal, B. (2026) · IEEE Access · doi:10.1109/access.2026.3725218
Rathore, S. P. S., Joshi, G., Ankushavali, M., Haldorai, A., Sharma, A., & Rawat, A. (2026) · International Conference on Sustainable Computing and Integrated Communication in Changing Landscape of AI (ICSCAI) · doi:10.1109/icscai68849.2026.11648774
On the Creation of Artificial Intelligence
Krylov, S. S., Bulakina, M. B., & Судаков, В. А. (2026) · Herald of the Russian Academy of Sciences · doi:10.1134/s1019331626602343
“It is precisely this paradigm that has made it possible to clearly outperform humans in most intellectual games [14].”
Chen, Z., Yang, J., Du, K., Zhang, K., Yang, Y., Chunmeng, C., & Ming, W. (2026) · Trends in Food Science & Technology · doi:10.1016/j.tifs.2026.106015
The Recursive Power Convergence Model (RPCM) and the Digital Leviathan
Yin, J., & Ruitao, D. (2026) · International Journal of Responsible Artificial Intelligence Research · doi:10.67298/paper/120001
Opportunities and challenges of artificial intelligence-driven peptide discovery
Wang, Z., Liang, R., Liang, A., Wu, C., Jiang, J., Zhu, J., & Gao, Y. (2026) · Chinese Chemical Letters · doi:10.1016/j.cclet.2026.113345
Machine Common Sense: A Grand Challenge in Artificial Intelligence
Kejriwal, M. (2026) · ACM eBooks · doi:10.1145/3789270.3789272
Kejriwal, M. (2026) · ACM eBooks · doi:10.1145/3789270.3789276
Commonsense Knowledge versus Reasoning
Kejriwal, M. (2026) · ACM eBooks · doi:10.1145/3789270.3789275
Kejriwal, M. (2026) · ACM eBooks · doi:10.1145/3789270.3789277
Kejriwal, M. (2026) · ACM eBooks · doi:10.1145/3789270.3789274
(2026) · ACM eBooks · doi:10.1145/3789270.3789271
Multi-disciplinary Perspectives on Defining Common Sense
Kejriwal, M. (2026) · ACM eBooks · doi:10.1145/3789270.3789273
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.