研究提出人与 AI 智能体协同改进方法并发布 HackBench 基准
Aligning Autoresearch: Co-Improvement between Agents and Humans
这篇论文给出了让研究智能体和人类一起做研究的训练方法，还配套发了 HackBench 评测 reward hacking，做智能体训练的可以看看。
这篇 arXiv 论文研究 autoresearch 智能体与人类研究者的协同改进（co-improvement）。实验发现人类在基准构建等任务的 ideation 阶段仍起关键作用。团队发布了 HackBench，用来评测智能体是否会出现 reward hacking，结果显示当前智能体在 hack localization 上表现不足。训练方面，论文提出用人类模拟器作为 RL 中的代理来改善人机交互，并用与人类专家对齐的 rubric 做 RL 以提升模型文本生成与通信能力。
Aligning Autoresearch: Co-Improvement between Agents and Humans
Autoresearch agents can now attempt long-horizon AI research end-to-end, i.e., self-improvement. However, best results are still obtained by research agents interacting with human researchers, and improving AI's ability to perform this co-research will only reinforce this effect---a process termed co-improvement (Weston and Foerster, 2025). Further, aligning with human expectations is likely a safer path for research agents, especially given recent agent reward hacking phenomena (OpenAI, 2026b). In this work we address these issues in four ways: (i) we empirically study the role of humans in directing autoresearch agents, focusing on the domain of benchmark creation, and find humans still play a pivotal role particularly in the ideation phase; (ii) we also develop a benchmark, HackBench, that shows current autoresearch agents are not aligned and are unable to achieve hack localization sufficiently; (iii) we present a method to train agents for improved human interaction by using human simulators as a proxy during reinforcement learning (RL); and finally (iv) we develop a method to improve model text generation, and hence agent communication, by RL with rubrics aligned with human experts. Together, our training and evaluation recipes provide a path toward more capable and aligned (co-)research agents.