Large language models instantiate evolutionarily robust strategies of cooperation
Saptarshi Pal, Abhishek Mallela, Lenz Pracher, Chiyu Wei, Feng Fu, Santiago Schnell, Martin A Nowak · 2026
Read Large language models instantiate evolutionarily robust strategies of cooperation by Saptarshi Pal, Abhishek Mallela, Lenz Pracher, Chiyu Wei, Feng Fu, Santiago Schnell, Martin A Nowak (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/PMC13298648.txt.
Description
Chatbots play the prisoner's dilemma in surprisingly cooperative ways, report Saptarshi Pal and colleagues in a 2026 study of five leading language models. The team probed how the models cooperate or defect across repeated rounds, varying payoffs, stopping chances, and whether the final round was known.
The models generally behaved generously, often landing on strategies that evolutionary theory deems robust, and they adjusted sensibly to many parameter changes. Yet none stayed fully consistent across every task, and some responses broke with both theory and human experimental patterns. Tournaments between the inferred strategies added further evidence. The authors frame the results as a check on trusting AI advice in social decisions.
Catalog record
| CREATOR | Saptarshi Pal, Abhishek Mallela, Lenz Pracher, Chiyu Wei, Feng Fu, Santiago Schnell, Martin A Nowak |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 54387 |
| LICENSE | https://creativecommons.org/licenses/by/4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |