Goto

Collaborating Authors

 cost and price tag


Playing Large Games with Oracles and AI Debate

arXiv.org Artificial Intelligence

We consider regret minimization in repeated games with a very large number of actions. Such games are inherent in the setting of AI safety via debate, and more generally games whose actions are language-based. Existing algorithms for online game playing require computation polynomial in the number of actions, which can be prohibitive for large games. We thus consider oracle-based algorithms, as oracles naturally model access to AI agents. With oracle access, we characterize when internal and external regret can be minimized efficiently. We give a novel efficient algorithm for internal regret minimization whose regret and computation complexity depend logarithmically on the number of actions. This implies efficient oracle-based computation of a correlated equilibrium in large games. We conclude with experiments in the setting of AI Safety via Debate that shows the benefit of insights from our algorithmic analysis.