Description
AI agents can sound cooperative while struggling to coordinate their work. Rohit Krishnan shares experiments with simulated towns and companies to examine how agents make decisions, respond to new information, and work together. In his tests, agents pursued narrow objectives, held onto outdated assumptions, and converged on groupthink in multiagent councils. He also found mismatches between confidence and performance. These observations lead to his proposal for environments where agents can practice judgment before taking on work at scale, alongside better tools for people to understand and direct their actions.
Follow O'Reilly on:
LinkedIn: https://www.linkedin.com/company/oreilly/
Facebook: http://facebook.com/OReilly
Instagram: https://www.instagram.com/oreillymedia
BlueSky: https://bsky.app/profile/oreilly.bsky.social
TikTok: https://www.tiktok.com/@oreillymedia