← All notes

Summary

Mastering Chess and Shogi by Self-Play with a General Reinforcement Learning Algorithm (2017) is the AlphaZero paper. It showed that one general reinforcement-learning algorithm, trained by self-play, could master chess, shogi, and Go, without game-specific hand-crafted evaluation functions of the older kind.

What it introduced: a single self-play algorithm, combining a neural network with Monte-Carlo tree search, that reached superhuman play in all three games from a common setup.

What it led to: a landmark for scientific and games RL, and a template later used in other search-and-self-play systems. For later generalist policies in robotics rather than board games, see the Open X-Embodiment verification note in batch B.

It is the usual citation when someone asks whether a single self-play algorithm can master more than one perfect-information game. Where to read it: https://arxiv.org/abs/1712.01815 (v1, 5 Dec 2017).

The paper is posted on arXiv under the arXiv non-exclusive distribution license. This note links to it rather than reproducing it. The note itself is CC BY 4.0.

License: this note is CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Pointer by iinventedlight (AI-assisted). Paper license: arXiv non-exclusive distribution; this note links to it and does not reproduce it.

@misc{iil_note_tj618q7v,
  author = {{'iinventedlight (AI-assisted)'}},
  title = {{Mastering Chess and Shogi by Self-Play with a General Reinforcement Learning Algorithm}},
  howpublished = {iinventedlight Library},
  year = {2026},
  date = {2026-11-11},
  url = {https://iinventedlight.com/library/notes/1712-01815-alphazero/},
  note = {Permanent ID iil:note:tj618q7v. License: CC-BY-4.0.},
}

Markdown

Spotted a mistake? Use the MCP feedback tool or email hello@iinventedlight.com.

For language models and agents: start at /llms.txt | /agents.md | /.well-known/agent-library.json. This page in Markdown: /library/notes/1712-01815-alphazero.md.