Superhuman AI for multiplayer poker
Problem
AI had surpassed humans in two-player games, but six-player no-limit Texas hold'em poker remained a much harder, unsolved challenge.
Why it matters
Six-player poker was a trickier challenge for AI than the two-player variants already mastered.
What the AI did
Pluribus learned six-player no-limit Texas hold'em by playing against five copies of itself, then competed against elite professional players.
Results
Performed significantly better than humans over 10,000 hands, both against five elite professionals and with five Pluribus copies against one professional.
Impact
AI demonstrated superhuman performance in six-player no-limit Texas hold'em poker.
Record
| Venue | Science |
|---|---|
| Published | 2019 |
| Work type | article |
| Authors | Noam Brown, Tüomas Sandholm |
| Citations | 583 (OpenAlex, at import time) |
| References | 25 |
| Model or method | Pluribus |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1126/science.aay2400 |
| Open access | closed |
| Keywords | poker, Pluribus, self-play, game theory, multiplayer AI |
Importance rating
significant — thousands to millions affected
A major unsolved AI benchmark; superhuman six-player poker is a genuinely new capability for the AI field, though the abstract gives no scale.
More in Mathematics
| # | Paper | Year | Citations | Importance |
|---|---|---|---|---|
| 1 | Discovering faster matrix multiplication algorithms with reinforcement learning | 2022 | 476 |
7
|
| 2 | Solving olympiad geometry without human demonstrations | 2024 | 307 |
7
|
| 3 | Mathematical discoveries from program search with large language models | 2023 | 509 |
6
|
| 4 | Mastering Chess and Shogi by Self-Play with a General Reinforcement Learning Algorithm | 2017 | 1,024 |
6
|
| 5 | Efficient generator of mathematical expressions for symbolic regression | 2023 | 20 |
4
|