Researchers at Duke University´s Fuqua School of Business have launched a new artificial intelligence-powered tool designed to predict the commercial value of scientific discoveries before these works garner citations or patents. This development, spotlighted in the journal ´Strategic Management Journal,´ introduces a methodology that enables firms, investors, and universities to identify research with high potential to drive economic innovation at very early stages.
The tool, called scientifiq.ai, leverages large language models and deep neural networks to analyze the abstracts of scientific papers. By translating text into numerical representations and training on a massive database of 139 million academic papers across natural and applied sciences from 2000 to 2020, the model predicts the likelihood that a given paper will eventually be cited in renewed patent applications—a proxy for commercial impact. In validation, articles ranked in the top quartile of scientifiq.ai’s measure were over 20 times more likely to be cited by a renewed patent than those in the bottom quartile.
This approach moves beyond traditional, backward-looking metrics such as citation counts or patent filings, which only reflect past scientific attention or commercial activity. Instead, scientifiq.ai offers universities an early signal to help plan licensing or commercialization strategies and gives companies a direct route to identifying academic experts who are unknowingly solving pertinent industry problems. The platform facilitates matches between resource-seeking organizations and researchers, helping break down barriers to tech transfer and leveling the playing field for promising work at lesser-known institutions. The underlying code has also been made available on GitHub, ensuring transparency and enabling other research teams to build upon the method. According to Fuqua’s Sharique Hasan, this tool aims to reduce bias in R&D selection and foster smarter, more inclusive innovation strategies by systematically surfacing hidden opportunities within the scientific community.
