Google DeepMind paper finds AI agent auctions reach 99.2% of ideal answer quality

Google DeepMind and Google Research Find AI Auctions Reach 99.2% of Oracle Welfare

In simulations on more than 30,000 prompts, 110 models reached 99.2% of an oracle allocation’s modeled welfare, measured using estimated correctness and a model-size cost proxy.

Google DeepMind paper finds AI agent auctions reach 99.2% of ideal answer quality

Google DeepMind researchers find first-price auctions for AI agents hit 99.2% of oracle welfare, beating the classic 63% worst-case bound.