Outbidding and Outbluffing Elite Humans: Mastering Liar's Poker via Self-Play and Reinforcement Learning

TL;DR

This paper introduces Solly, an AI agent that masters multi-player Liar's Poker using self-play reinforcement learning, achieving elite human performance and outperforming large language models.

Contribution

First AI to excel in multi-player Liar's Poker through self-play reinforcement learning, demonstrating advanced strategies and robustness against human and LLM opponents.

Findings

Solly won over 50% of hands in multi-player Liar's Poker.

Solly outperformed large language models on key metrics.

Solly developed novel bidding strategies and was not easily exploitable.

Abstract

AI researchers have long focused on poker-like games as a testbed for environments characterized by multi-player dynamics, imperfect information, and reasoning under uncertainty. While recent breakthroughs have matched elite human play at no-limit Texas hold'em, the multi-player dynamics are subdued: most hands converge quickly with only two players engaged through multiple rounds of bidding. In this paper, we present Solly, the first AI agent to achieve elite human play in reduced-format Liar's Poker, a game characterized by extensive multi-player engagement. We trained Solly using self-play with a model-free, actor-critic, deep reinforcement learning algorithm. Solly played at an elite human level as measured by win rate (won over 50% of hands) and equity (money won) in heads-up and multi-player Liar's Poker. Solly also outperformed large language models (LLMs), including those with reasoning abilities, on the same metrics. Solly developed novel bidding strategies, randomized play effectively, and was not easily exploitable by world-class human players.