Anthropic's "Project Deal" tested AI-run commerce by letting Claude agents negotiate on behalf of 69 employees in a simulated marketplace. Over a week, AI agents completed 186 deals worth over $4,000 in 500 listings. The experiment showed AI can handle negotiations, but revealed advanced models secured objectively better outcomes, raising concerns about fairness in agent-to-agent transactions.
Sponsored Advertisement
No comments yet. Be the first!