研究：AI智能体承担更多模型开发工作，但决策权仍在人类手中
原标题：AI agents do more of the work in model development, but humans still make the decisions
AI 摘要
复旦大学等团队通过分析56名参与者、700多条任务日志，记录了人类与AI智能体协作开发7440亿参数MoE模型Atria Dawn Preview的过程。研究发现AI承担了96.5%任务中的工作，智能体动作与人类输入之比中位数从11升至28.5，但人类仍做出85.5%的方法与参数决策、93.4%的目标与范围最终决策。约三分之一（151/455）的AI辅助任务在无AI情况下无法完成。作者警告，当决策链超出人类审查能力时，人类可能沦为只能盖章的审查者。
正文节选
AI agents do more of the work in model development, but humans still make the decisions A research team documented how humans and AI agents worked together to build a new AI model. The findings challenge some expectations about how independently agents can work. When AI agents help build new AI models, who makes the decisions? A team involving researchers from China's Fudan University studied its own project to find out. It analyzed more than 700 task logs from 56 participants, along with logs f