External information environment required for cross disciplinary thinking in large language models based on interdisciplinary lay summary writing
Year: 2025
Type: article
Abstract: External information stimuli are crucial for humans to reintegrate knowledge and activate cross-disciplinary thinking. This study investigates how the external information environment affects LLMs' cross-disciplinary thinking through analyzing generated interdisciplinary lay summaries. The interdisciplinary lay summary, building on traditional lay summaries for non-specialists, integrates intersection interpretation and cross-inspiration to assess convergent and divergent thinking. This study utilizes general abstracts and lay summaries to create contexts of deep knowledge and broad knowledge. By comparing intersection interpretations and cross-inspirations generated by LLMs under different information input combinations, we explore the external information environment's influence on cross-disciplinary thinking. ChatGPT-4.0 and Claude3 are used in this study. Findings indicate that simultaneous access to deep and broad knowledge enhances LLMs' convergent thinking, while intersection interpretations foster divergent thinking. Furthermore, a modest correlation exists between the external information environment and divergent thinking. This study advances the LLMs' application in cross-disciplinary research.
Source: Scientific Reports
Authors: Mei Chen, Yijun Su, Junyuan Guo, Xiaoying Liu
Institution: Minzu University of China
Language: English
Topic: Topic Modeling
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $2,690
Award: G6596213355
OpenAlex ID: W4415992473