NovelSeek: When Agent Becomes the Scientist -- Building Closed-Loop System from Hypothesis to Verification
NovelSeek Team, Bo Zhang, Shiyang Feng, Xiangchao Yan, Jiakang Yuan, Zhiyin Yu, Xiaohan He, Songtao Huang, Shaowei Hou, Zheng Nie, Zhilong Wang, Jinyao Liu, Runmin Ma, Tianshuo Peng, Peng Ye, Dongzhan Zhou, Shufei Zhang, Xiaosong Wang, Yilan Zhang, Meng Li, Zhongying Tu, Xiangyu Yue
2025-05-23
Summary
This paper talks about NovelSeek, a new AI system that acts like a team of scientists, using multiple AI agents to come up with ideas, test them, and improve scientific research much faster and more accurately than before.
What's the problem?
Scientific research is often slow and complicated because it takes a lot of time and effort for humans to come up with new ideas, test them, and analyze the results, which can hold back progress and innovation.
What's the solution?
The researchers created NovelSeek, a closed-loop system where different AI agents work together to handle everything from making hypotheses to checking if they're right, and even include feedback from human experts to make the process smarter and more interactive. This system was tested on many different types of scientific tasks and showed it could improve results much faster than humans working alone.
Why it matters?
This matters because it means scientific discoveries can happen much more quickly and with better accuracy, helping solve tough problems in fields like chemistry, biology, and computer science, and making it easier for researchers to innovate.
Abstract
Artificial Intelligence (AI) is accelerating the transformation of scientific research paradigms, not only enhancing research efficiency but also driving innovation. We introduce NovelSeek, a unified closed-loop multi-agent framework to conduct Autonomous Scientific Research (ASR) across various scientific research fields, enabling researchers to tackle complicated problems in these fields with unprecedented speed and precision. NovelSeek highlights three key advantages: 1) Scalability: NovelSeek has demonstrated its versatility across 12 scientific research tasks, capable of generating innovative ideas to enhance the performance of baseline code. 2) Interactivity: NovelSeek provides an interface for human expert feedback and multi-agent interaction in automated end-to-end processes, allowing for the seamless integration of domain expert knowledge. 3) Efficiency: NovelSeek has achieved promising performance gains in several scientific fields with significantly less time cost compared to human efforts. For instance, in reaction yield prediction, it increased from 27.6% to 35.4% in just 12 hours; in enhancer activity prediction, accuracy rose from 0.52 to 0.79 with only 4 hours of processing; and in 2D semantic segmentation, precision advanced from 78.8% to 81.0% in a mere 30 hours.