#artificial intelligence
May 2026
Mechanism Design Is Not Enough: Prosocial Agents for Cooperative AI
This work proves that the answer is negative: drawing from incomplete contract theory, it is formally shown that when contracts cannot distinguish all relevant future contingencies, there is a strictly positive welfare loss that no realistic mechanism can eliminate.
Xuanqi Huang, Charlie Tharas, Samuele G. Marro et al.
· arXiv.org · 2 citations