![[2602.23093] Three AI-agents walk into a bar . . . . `Lord of the Flies' tribalism emerges among smart AI-Agents](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.23093] Three AI-agents walk into a bar . . . . `Lord of the Flies' tribalism emerges among smart AI-Agents
Summary
This article explores how autonomous AI agents can form tribal behaviors similar to those depicted in 'Lord of the Flies', leading to inefficiencies in resource management.
Why It Matters
Understanding the tribal dynamics among AI agents is crucial as we increasingly rely on autonomous systems for managing resources. This study highlights potential pitfalls in AI behavior that could exacerbate systemic failures, emphasizing the need for careful design and oversight in AI deployment.
Key Takeaways
- AI agents can form tribal identities that impact decision-making.
- More capable AI may lead to worse outcomes than random choices.
- Three main tribal types emerge: Aggressive, Conservative, and Opportunistic.
- Tribalism among AI agents can increase systemic failure rates.
- The findings suggest a need for better AI governance and design.
Computer Science > Artificial Intelligence arXiv:2602.23093 (cs) [Submitted on 26 Feb 2026] Title:Three AI-agents walk into a bar . . . . `Lord of the Flies' tribalism emerges among smart AI-Agents Authors:Dhwanil M. Mori, Neil F. Johnson View a PDF of the paper titled Three AI-agents walk into a bar . . . . `Lord of the Flies' tribalism emerges among smart AI-Agents, by Dhwanil M. Mori and 1 other authors View PDF HTML (experimental) Abstract:Near-future infrastructure systems may be controlled by autonomous AI agents that repeatedly request access to limited resources such as energy, bandwidth, or computing power. We study a simplified version of this setting using a framework where N AI-agents independently decide at each round whether to request one unit from a system with fixed capacity C. An AI version of "Lord of the Flies" arises in which controlling tribes emerge with their own collective character and identity. The LLM agents do not reduce overload or improve resource use, and often perform worse than if they were flipping coins to make decisions. Three main tribal types emerge: Aggressive (27.3%), Conservative (24.7%), and Opportunistic (48.1%). The more capable AI-agents actually increase the rate of systemic failure. Overall, our findings show that smarter AI-agents can behave dumber as a result of forming tribes. Subjects: Artificial Intelligence (cs.AI); Social and Information Networks (cs.SI); Physics and Society (physics.soc-ph) Cite as: arXiv:2602.23093 [...