Resource
AI in Science: Evidence of impact from AlphaFold 2
Artificial intelligence tools and methods are diffusing within scientific research. Due to their emerging nature, questions exist over the potential benefits and drawbacks they present for the productivity, diversification and quality of research. However, robust evidence of systemic impacts of AI in science is limited.
We aim to address this gap by analysing the impact of AlphaFold 2, an AI tool developed by Google DeepMind in 2021, that addresses a longstanding problem in structural biology – protein structure prediction. AlphaFold 2 presents a useful case study of AI in science due to the accuracy of its predictions, its unanticipated development, and its free and open availability.
In this report, we study and compare the impact of AlphaFold 2 against typical structural biology research, and other high-impact, contemporary developments across four key dimensions: scientific reach, experimental structural biology, academic productivity and quality, and applied research and innovation.
Our findings from this work are split into two outputs:
Summary report: A high-level overview of our approach and key findings, with conclusions and recommendations.
Full report: (Coming on 28/11/25) A full-length report with a comprehensive literature review, methodology, results and discussion, as well as an extended appendix.
We present this report as an independent analysis of the scientific impact of AlphaFold 2, and a contribution to the evidence base around AI in science. Our findings are not conclusive, and many questions remain in this key policy area to be answered in future work.
Scientific reach
We assess the influence and diffusion of AlphaFold 2 and comparable frontier developments across the publication and researcher ecosystem.
Experimental structural biology
We explore AlphaFold’s contribution to experimental structural biology, examining novel submissions to the Protein DataBank.
Academic productivity and quality
We investigate whether AlphaFold 2 and other frontier works support researchers to publish additional publications, with higher quality.
Applied research and innovation
We examine links between works that build on AlphaFold 2 and applied research outputs, including patents, clinical studies and disease related research.
Project team
-
David Ampudia
Senior Data Scientist -
George Richardson
Head of Data Science and Technology
IGL acknowledges support from Google DeepMind to carry out this work as an independent assessment of the impact of AlphaFold 2. IGL retained full control of the methodology, analysis and report.