Suspicion-Agent

Overview

Suspicion-Agent is a Python implementation of a research paper that uses theory of mind reasoning with GPT-4 to play imperfect information games. It models other players' beliefs and intentions to make strategic decisions in games like Avalon.

Best for

Best for
Researchers and developers exploring theory of mind reasoning in multi-agent game settings

Use cases

• Building game agents that reason about opponents' hidden knowledge
• Testing theory of mind capabilities in large language models
• Researching multi-agent strategic interaction with incomplete information

Notes

Suspicion-Agent is a Python implementation of a research paper that uses theory of mind reasoning with GPT-4 to play imperfect information games. It models other players’ beliefs and intentions to make strategic decisions in games like Avalon.

165 stars on GitHub. Last updated 2023-11-08.

Use cases

• Building game agents that reason about opponents’ hidden knowledge
• Testing theory of mind capabilities in large language models
• Researching multi-agent strategic interaction with incomplete information

Pros

• Implements a novel theory of mind approach for imperfect information games
• Leverages GPT-4’s reasoning for strategic decision making
• Open source with 165 GitHub stars and active community interest

Cons

• Requires GPT-4 API access which incurs ongoing costs
• Limited to specific game environments and may not generalize easily
• Research prototype not production ready for general use

Indexed from awesome-ai-agents and enriched against its public facts.