1.【讲座】

Science Meets LLMs: Integrating Large Language Models into Scientific Discovery

科学与大语言模型的交融：将大语言模型融入科学探究

2.【主讲人】

下川敏雄 教授 和歌山县立医科大学

3.【主讲人介绍】

Dr. Toshio Shimokawa is Professor of Medical Statistics at Wakayama Medical University, Director of its Clinical Research Center and Data Science Center. His research covers statistical machine learning for medical data, with more than 300 peer-reviewed publications. He serves as visiting researcher and statistical advisor for Japanese public institutions, and holds board positions in several academic societies.

下川敏雄，现任和歌山医科大学医学统计学教授，临床研究中心、数据科学中心主任。主攻面向医学数据的统计机器学习，发表同行评审论文300余篇。兼任日本多家公共机构客座研究员与统计顾问。

4.【讲座内容简介】

This lecture examines how large language models (LLMs) can support scientific discovery while addressing fabricated references, unreliable analyses, and overlooked inconsistencies. It emphasizes that effective use requires clear research questions, defined concepts, and explicit verification procedures, rather than prompt-writing techniques alone.The lecture covers research design, conducting research, paper writing, and presentation preparation. It distinguishes decisions researchers must make from tasks that can be delegated to LLMs, including analytical support, drafting, translation, and slide preparation. Researchers must verify outputs against data and research logic, treating LLMs as partners in thinking while retaining final judgment and responsibility.

本讲座探讨了大语言模型（LLMs）如何在支持科学发现的同时，应对虚假引用、不可靠分析以及被忽视的矛盾问题。讲座强调，有效使用大语言模型需要明确的研究问题、清晰的定义概念以及明确的验证流程，而不能仅依赖于快速写作技巧。讲座涵盖了研究设计、研究实施、论文撰写以及演示文稿准备等内容。它区分了研究人员必须做出的决策与可委托给大语言模型的任务，包括数据分析支持、草拟、翻译和幻灯片制作等。研究人员需将输出结果与数据及研究逻辑进行核对，应将大语言模型视为思维伙伴，同时保留最终判断和责任。

5.【时间】

2026年11月13日（周五）16:00-17:00

6.【地点】

西南交通大学犀浦校区11号教学楼交运大楼304会议室