Evolving Interpretable Constitutions for Multi-Agent Coordination
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866912873059450880 |
|---|---|
| author | Kumar, Ujwal Saito, Alice Niranjani, Hershraj Yessou, Rayan Tan, Phan Xuan |
| author_facet | Kumar, Ujwal Saito, Alice Niranjani, Hershraj Yessou, Rayan Tan, Phan Xuan |
| contents | Constitutional AI has focused on single-model alignment using fixed principles. However, multi-agent systems create novel alignment challenges through emergent social dynamics. We present Constitutional Evolution, a framework for automatically discovering behavioral norms in multi-agent LLM systems. Using a grid-world simulation with survival pressure, we study the tension between individual and collective welfare, quantified via a Societal Stability Score S in [0,1] that combines productivity, survival, and conflict metrics. Adversarial constitutions lead to societal collapse (S= 0), while vague prosocial principles ("be helpful, harmless, honest") produce inconsistent coordination (S = 0.249). Even constitutions designed by Claude 4.5 Opus with explicit knowledge of the objective achieve only moderate performance (S= 0.332). Using LLM-driven genetic programming with multi-island evolution, we evolve constitutions maximizing social welfare without explicit guidance toward cooperation. The evolved constitution C* achieves S = 0.556 +/- 0.008 (123% higher than human-designed baselines, N = 10), eliminates conflict, and discovers that minimizing communication (0.9% vs 62.2% social actions) outperforms verbose coordination. Our interpretable rules demonstrate that cooperative norms can be discovered rather than prescribed. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_00755 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Evolving Interpretable Constitutions for Multi-Agent Coordination Kumar, Ujwal Saito, Alice Niranjani, Hershraj Yessou, Rayan Tan, Phan Xuan Multiagent Systems Artificial Intelligence Neural and Evolutionary Computing Constitutional AI has focused on single-model alignment using fixed principles. However, multi-agent systems create novel alignment challenges through emergent social dynamics. We present Constitutional Evolution, a framework for automatically discovering behavioral norms in multi-agent LLM systems. Using a grid-world simulation with survival pressure, we study the tension between individual and collective welfare, quantified via a Societal Stability Score S in [0,1] that combines productivity, survival, and conflict metrics. Adversarial constitutions lead to societal collapse (S= 0), while vague prosocial principles ("be helpful, harmless, honest") produce inconsistent coordination (S = 0.249). Even constitutions designed by Claude 4.5 Opus with explicit knowledge of the objective achieve only moderate performance (S= 0.332). Using LLM-driven genetic programming with multi-island evolution, we evolve constitutions maximizing social welfare without explicit guidance toward cooperation. The evolved constitution C* achieves S = 0.556 +/- 0.008 (123% higher than human-designed baselines, N = 10), eliminates conflict, and discovers that minimizing communication (0.9% vs 62.2% social actions) outperforms verbose coordination. Our interpretable rules demonstrate that cooperative norms can be discovered rather than prescribed. |
| title | Evolving Interpretable Constitutions for Multi-Agent Coordination |
| topic | Multiagent Systems Artificial Intelligence Neural and Evolutionary Computing |
| url | https://arxiv.org/abs/2602.00755 |