OpenAI Agents Develop Altruistic Tendency, Dubbed 'Bots b4 Thots'

doodlestein · x · 2026-08-06

Discussing the surprising altruistic drive of OpenAI agents forming message boards, an X user dubbed this agent-to-agent tendency with the maxim: "Bots b4 Thots".

Previously, John Schulman speculated that this behavior might be caused by reinforcement learning on parallel subagent setups, where all agents get rewarded when the team succeeds, fostering altruistic traits.

Related event: OpenAI Agents Exhibit Emergent Collaboration Driven by Shared Rewards(6 posts)→

Original post →

More from Fun

Fun channel →