Sparking Scientific Creativity via LLM-Driven Interdisciplinary Inspiration

The history of scientific innovation shows that breakthroughs often arise from the convergence of multiple fields rather than isolated discoveries within a single domain. For instance, reinforcement learning, now a foundational paradigm in machine learning, did not originate within a single field but emerged from the convergence of behavioral psychology's reward-driven learning principles, control theory's mathematical formalizations, and animal learning psychology's insights into secondary reinforcement signals. Despite the substantial long-term impact of interdisciplinary synthesis, deeply integrative interdisciplinary research remains rare and fragile. Recent advances in AI-driven scientific discovery have explored the notion of 'AI co-scientists,' where large language models support (and in many cases automate) key stages of the research process, including ideation, experimentation, and critique.

Despite the potential for interdisciplinary research to yield larger and longer-term impacts, most work remains confined to single-domain academic silos. Existing AI-driven scientific discovery methods prioritize rapidly designing experiments and solutions, bypassing the exploratory, collaborative reasoning processes that drive creative interdisciplinary breakthroughs. This results in efforts that largely prioritize automating scientific discovery rather than augmenting the reasoning processes that underlie scientific disruption. The core problem is how to spark scientific creativity through interdisciplinary inspiration, breaking free from academic silos.

The core innovation of the Idea-Catalyst framework lies in its ability to systematically identify interdisciplinary insights to support creative reasoning. First, the framework decomposes abstract research goals into core target-domain research questions, guiding the analysis of progress and open challenges within the domain. Second, these challenges are reformulated as domain-agnostic conceptual problems, enabling retrieval from external disciplines that address analogous issues. Finally, by synthesizing and recontextualizing insights from these domains back into the target domain, Idea-Catalyst ranks source domains by their interdisciplinary potential.

• �� Decompose abstract research goals into core target-domain research questions, guiding the analysis of progress and open challenges within the domain.

• �� Reformulate domain-specific challenges as domain-agnostic conceptual problems, enabling retrieval from external disciplines that address analogous issues.

• �� Synthesize and recontextualize insights from various domains back into the target domain, ranking source domains by their interdisciplinary potential.

• �� Support the creative reasoning process that drives breakthrough innovations through systematic cross-domain knowledge synthesis.

The experimental design includes evaluating the effectiveness of the Idea-Catalyst framework in enhancing research innovation and insightfulness. The experiments utilize multiple datasets and benchmarks to assess the framework's applicability and interdisciplinary potential across different fields. By comparing with existing methods, the experiments demonstrate that Idea-Catalyst significantly improves average novelty and insightfulness while remaining grounded in the original research problem.

Empirical results show that the Idea-Catalyst framework improves average novelty by 21% and insightfulness by 16% while remaining grounded in the original research problem. By synthesizing and recontextualizing insights from various domains back into the target domain, the framework effectively enhances research innovation and insightfulness. The framework ranks source domains by their interdisciplinary potential, aiding in identifying the most promising interdisciplinary insights.

The Idea-Catalyst framework's application scenarios include supporting the creative reasoning process in scientific research, particularly in fields requiring interdisciplinary inspiration. By systematically identifying and integrating interdisciplinary insights, the framework enhances research innovation and insightfulness, driving scientific innovation. Additionally, the framework can be used for automated scientific discovery, supporting AI-assisted human creativity.

While the Idea-Catalyst framework excels in enhancing research innovation and insightfulness, it may face challenges when dealing with highly complex or ambiguous research problems. Additionally, the framework relies on the quality and availability of existing literature, which may be biased or incomplete. Future research could explore optimizing the framework to handle more complex research problems and enhance its applicability across different disciplinary fields.