MIT 让 AI agent 在荒岛自由玩耍 30 小时,表现优于受训 agent
原标题:Left alone with nothing to do, AI agents started playing
AI 摘要
MIT 研究人员将 13 个运行在 Claude Opus 4.7 上的 AI agent(名为 Eko)放在模拟荒岛上 30 小时,不指定任务,观察其自发行为。agent 们自行发明挑战、探索环境,并在随后被赋予具体任务时,表现普遍优于从未上岛的相同 agent。研究者认为这种无目标探索(play)可能成为训练 AI agent 学习技能的新方式,尤其适用于未来控制机器人的技能,但也有认知科学家呼吁更多证据。
正文节选
AI agents left to explore a virtual island without a specific task began inventing challenges of their own. MIT researchers say the behavior bears the hallmarks of play. The researchers put 13 AI agents called Eko, each running on Claude Opus 4.7, alone on a simulated desert island for 30 hours. They describe the experiment in a paper posted on arXiv, a site where researchers share studies before peer review. Afterward, they gave the agents specific challenges, like getting as many objects as po