ScienceBuddy: Recursive-in-Recursive Self-Improvement for Interactive Scientific Agents

介绍

AI 科研搭档

可交互的 Scientific Agent：关注真实科研交互中的递归式自我改进，重点探索 Recursive-in-Recursive Self-Improvement（RSI in RSI）。

ScienceBuddy 是一项围绕科学家交互、科研反馈和智能体持续改进展开的独立研究与工程探索，与 DFM 所关注的 Discovery Intelligence 方向一致。

它重点探索一种交互式 Science RSI 形态：用户提出科学问题，智能体调用工具并给出回应；科学家的追问、修改、采纳、否定和重新运行形成反馈，进而推动科研智能体工作环境和 Harness 的持续演化。

ScienceBuddy 的产出不只是一次性的问答结果，也包括交互过程中的反馈、研究轨迹、工具调用和任务状态。这些信息未来可以进一步整理为可复用的科学任务和环境，用于强化学习或其他模型训练。

承接并结构化

科学问题
数据需求
研究材料
文献和代码
科研工具调用
假设形成与验证
后续科研工作流所需的信息

当前为早期产品 / 研究预览形态，具体开放能力、演示方式和使用入口以实际版本为准。产品入口于 9 月 16 日开放。

摘要

ScienceBuddy is an interactive scientific research workspace that brings continually improving agents into a researcher's everyday work. It supports researchers in carrying out scientific tasks while turning their requests, feedback and execution evidence into tasks and evaluation rubrics for continual learning. At its core is recursive-in-recursive self-improvement, which couples harness evolution with model reinforcement learning: the inner recursion improves the harness with the model fixed, and the outer recursion trains the model under the improved harness. Case studies cover researcher interaction, harness refinement and model learning across four families of scientific task.

引用

BibTeX

@article{xue2026sciencebuddy,
  title={ScienceBuddy: Recursive-in-Recursive Self-Improvement for Interactive Scientific Agents},
  author={Xue, Shuhan and Zhong, Jianyuan and Nan, Ziyuan and Li, Wenbin and Yu, Zhaochen and Ding, Jinchao and Gao, Qiang and Zhan, Pengyu and Zhang, Yuntong and Cheng, Tian and Yin, Zhenfei and Wu, Yingcheng and Yang, Ling},
  journal={arXiv preprint arXiv:2609.17523},
  year={2026}
}

ScienceBuddy: Recursive-in-Recursive Self-Improvement for Interactive Scientific Agents

介绍

摘要

作者

期刊/会议

引用