LLM Agents as AI Scientists: A Survey
Keywords: automated research, AI4Science
Abstract: In recent years, the rapid development of Large Language Models (LLMs) has significantly reshaped the landscape of scientific research, providing powerful support throughout the research lifecycle. In this survey, we investigate current research concerning the transformative impact of agentic LLMs on the scientific process. Starting from an analysis of the limitations of human research, we examine the distinct contributions of LLMs across different key stages: hypothesis discovery, experiment implementation, paper writing, and peer reviewing. Our analysis highlights task-specific approaches and evaluation standards from a human-centered perspective, offering a detailed overview of the current state of the field. By outlining existing challenges and future directions, we hope this serves as a guide for researchers and practitioners seeking to harness these models to propel scientific discovery.
Submission Number: 1
Loading