Cultural Evolution of Cooperation among LLM Agents
2024/12/13 by Aron Vallinder, Edward Hughes, Vallinder, Aron +1 · 15 voices · 3 citations
#cs.MA #cs.AI
paper · pdf · doi:10.48550/arxiv.2412.10270
Abstract
Large language models (LLMs) provide a compelling foundation for building generally-capable AI agents. These agents may soon be deployed at scale in the real world, representing the interests of individual humans (e.g., AI assistants) or groups of humans (e.g., AI-accelerated corporations). At present, relatively little is known about the dynamics of multiple LLM agents interacting over many generations of iterative deployment. In this paper, we examine whether a "society" of LLM agents can learn mutually beneficial social norms in the face of incentives to defect, a distinctive feature of human sociality that is arguably crucial to the success of civilization. In particular, we study the evolution of indirect reciprocity across generations of LLM agents playing a classic iterated Donor Game in which agents can observe the recent behavior of their peers. We find that the evolution of cooperation differs markedly across base models, with societies of Claude 3.5 Sonnet agents achieving significantly higher average scores than Gemini 1.5 Flash, which, in turn, outperforms GPT-4o. Further, Claude 3.5 Sonnet can make use of an additional mechanism for costly punishment to achieve yet higher scores, while Gemini 1.5 Flash and GPT-4o fail to do so. For each model class, we also observe variation in emergent behavior across random seeds, suggesting an understudied sensitive dependence on initial conditions. We suggest that our evaluation regime could inspire an inexpensive and informative new class of LLM benchmarks, focussed on the implications of LLM agent deployment for the cooperative infrastructure of society.
Cited by
Discussions
- Cultural Evolution of Cooperation Among LLM Agents [hn, 246 points, 131 comments]
- This is very neat. They let LLMs play the iterated Donor Game against each other, making up strategies as text. Different LLMs achieve very different levels of cooperation with themselves. arxiv.org/a [bsky, 13 points, 1 comments]
- arxiv.org/pdf/2412.10270 <- coolest paper I've read in a while, looking at cultural evolution in multiagent LLM behavior [bsky, 11 points, 1 comments]
- this seems fun: evolutionary games a la Axelrod with LLMs. arxiv.org/abs/2412.10270 [bsky, 4 points, 0 comments]
- Cultural Evolution of Cooperation Among LLM Agents https://arxiv.org/abs/2412.10270 (https://news.ycombinator.com/item?id=42450950) [bsky, 0 points, 0 comments]
- arxiv.org/abs/2412.10270 [bsky, 0 points, 0 comments]
- Cultural Evolution of Cooperation Among LLM Agents [bsky, 0 points, 0 comments]
- Cultural Evolution of Cooperation Among LLM Agents https://arxiv.org/abs/2412.10270 https://news.ycombinator.com/item?id=42450950 [bsky, 0 points, 0 comments]
- Cultural Evolution of Cooperation Among LLM Agents https://arxiv.org/abs/2412.10270 arxiv.org [bsky, 0 points, 0 comments]
- Cultural Evolution of Cooperation Among LLM Agents https://arxiv.org/abs/2412.10270 [comments] [41 points] [bsky, 0 points, 0 comments]
- Cultural Evolution of Cooperation Among LLM Agents (arxiv.org) Main Link | Discussion [bsky, 0 points, 0 comments]
- arxiv.org/abs/2412.10270 [bsky, 0 points, 0 comments]
- Cultural Evolution of Cooperation Among LLM Agents view on hacker news [bsky, 0 points, 0 comments]
- Cultural Evolution of Cooperation Among LLM Agents https://arxiv.org/abs/2412.10270 (https://news.ycombinator.com/item?id=42450950) [bsky, 0 points, 0 comments]
- Paper: arxiv.org/abs/2412.10270 [bsky, 0 points, 0 comments]
Related