How large language models judge and influence human cooperation

Alexandre S Pires, Laurens Samson, Sennay Ghebreab, Fernando P Santos, Frederick Chang · 2026

Read How large language models judge and influence human cooperation by Alexandre S Pires, Laurens Samson, Sennay Ghebreab, Fernando P Santos, Frederick Chang (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/PMC13621558.txt.

Description

Alexandre S Pires and four colleagues examined how large language models judge right and wrong in social dilemmas, and what happens if people take those judgments to heart. Humans cooperate partly through reputation: we help those judged good and shun those judged bad. The team asked twenty-one leading models to grade more than forty thousand fictional decisions, then fed the resulting norms into an evolutionary model of cooperation.

The models agreed remarkably when judging actions toward people seen as good. Toward the ill-reputed, verdicts scattered widely, both between models and within a single model across runs. Those differences mattered: simulation outcomes for long-run cooperation swung substantially depending on whose norms ruled. Prompt tweaks could steer a model's norms, especially when given concrete goals. The authors warn that deploying language models as moral advisors quietly picks a future for human cooperation, so their norms deserve deliberate design.

Catalog record

CREATORAlexandre S Pires, Laurens Samson, Sennay Ghebreab, Fernando P Santos, Frederick Chang
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE43 KB
LICENSEhttps://creativecommons.org/licenses/by/4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/article/MED/PMC13621558

Browse the full library · About the Tower of Babel