Skip to content

Source paper

AI Can Learn Scientific Taste

Jingqi Tong, Mingzhe Li, Hangcheng Li and 20 more

Jingqi Tong, Mingzhe Li, Hangcheng Li, Yongzhuo Yang, Yurong Mou, Weijie Ma, Zhiheng Xi, Hongji Chen, Xiaoran Liu, Qinyuan Cheng, Ming Zhang, Qiguang Chen, Weifeng Ge, Qipeng Guo, Tianlei Ying, Tianxiang Sun, Yining Zheng, Xinchi Chen, Jun Zhao, Ning Ding, Xuanjing Huang, Yugang Jiang, Xipeng Qiu

arXiv 2603.14473 · 2026220 upvotes on Hugging Face

Research ideas from this paper (3)

Generated by AI. The paper’s authors did not write these.

AI summary

This paper explores whether AI models can develop 'scientific taste' to evaluate and propose high-impact research ideas. The authors propose Reinforcement Learning from Community Feedback (RLCF) to train models using large-scale citation data, successfully creating an AI that generates research proposals with higher potential impact than baseline models.

How it works (AI summary)

The methodology utilizes a dataset of field- and time-matched paper pairs (SciJudgeBench) with citation counts as preference labels. A generative reward model (Scientific Judge) is trained via Group Relative Policy Optimization (GRPO) to predict higher-cited papers, which then supervises a policy model (Scientific Thinker) via Comparison-Based GRPO to generate high-impact follow-up research ideas.