External information environment required for cross disciplinary thinking in large language models based on interdisciplinary lay summary writing.

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.