ScienceBuddy builds smarter scientific assistants that improve themselves
ScienceBuddy: Recursive-in-Recursive Self-Improvement for Interactive Scientific Agents
Artificial IntelligenceComputation and Language
Summary
Scientists often need help managing many steps in their research. This paper introduces ScienceBuddy, a tool that works alongside researchers to assist with scientific tasks while learning from feedback and results. It uses a special method called recursive-in-recursive self-improvement, where the tool and its rules get better by taking turns improving each other. This way, ScienceBuddy can continually become more helpful as it works with scientists on different projects.
What this means in practice
- •For software engineers: Integrate interactive AI assistants that improve through user feedback for complex task workflows.
- •For data science teams: Use adaptive AI systems that evolve task evaluation methods and improve model performance with ongoing data and execution evidence.
Authors
Shuhan Xue, Jianyuan Zhong, Ziyuan Nan, Wenbin Li, Zhaochen Yu, Jinchao Ding, Qiang Gao, Pengyu Zhan, Yuntong Zhang, Tian Cheng, Zhenfei Yin, Yingcheng Wu, Ling Yang
Abstract
We introduce and release ScienceBuddy, an interactive scientific research workspace that brings continually improving scientific agents into researchers' everyday workflows. ScienceBuddy supports researchers in carrying out scientific tasks while transforming their requests, feedback, and execution evidence into tasks and evaluation rubrics for continual learning. At its core is recursive-in-recursive self-improvement, a paradigm that couples harness evolution with model reinforcement learning: the inner recursion improves the harness with the model fixed, while the outer recursion trains the model under the improved harness. Harness evolution shapes training experience, and model learning creates new opportunities for harness adaptation. We present case studies of researcher interaction, harness refinement, and model learning, with the benchmark cases spanning four scientific task families. By releasing ScienceBuddy as a research product, we make this paradigm available to the scientific community and take a step toward discovery intelligence: scientific AI that advances through sustained collaboration with researchers and evolves alongside the research it supports. Website: http://science-buddy.io