Knowledge flows from science to AI technology: Identifying core and brokerage technological roles

Area of Science:

• Artificial Intelligence (AI)
• Innovation Studies
• Bibliometrics

Background:

• Artificial intelligence (AI) is rapidly advancing and integrating across industries.
• The flow of knowledge from scientific research to AI technological development is crucial but understudied.
• Existing research lacks a systematic analysis of AI's science-to-technology knowledge pathways.

Purpose of the Study:

• To systematically investigate the science-to-technology knowledge flow underpinning AI's evolution.
• To propose and apply a semantic framework for exploring AI-related knowledge transfer.
• To identify key technological trends and structural pathways in AI development.

Main Methods:

• Developed a two-stage framework: technology classification and semantic topic exploration.
• Classified AI patents using centrality measures from a CPC co-occurrence network.
• Applied BERTopic modeling to patent and scientific publication abstracts, using generative AI for topic labeling.

Main Results:

• Analyzed AI patents filed between 2002 and 2021.
• Traced key technological trends and identified structural pathways of knowledge flow.
• Revealed the dynamic interplay between scientific discovery and AI innovation.

Conclusions:

• The findings provide a structural understanding of AI's technological evolution driven by scientific knowledge.
• Offers practical implications for corporate research and development (R&D) strategies.
• Informs innovation policy design for the rapidly evolving AI landscape.