Google launches Gemini for Science, three experimental AI research tools
Google announced Gemini for Science at its I/O conference on May 21, offering researchers access to experimental AI tools designed for literature analysis, hypothesis development, and computational modeling. The tools will roll out gradually through Google Labs starting immediately.
The suite includes three applications: Literature Insights searches scientific papers and structures findings into data tables, reports, and presentations. Hypothesis Generation uses a multi-agent system to help researchers define challenges, generate ideas, and evaluate research plans with cited evidence. Computational Discovery tests code variations in parallel for applications like solar forecasting and disease modeling.
Google also bundled Science Skills into its Antigravity platform, integrating data from more than 30 life science databases including UniProt and AlphaFold Database. Early testing showed the bundle completed analyses in minutes that typically take hours.
Who's already using these tools
More than 100 academic institutions are validating the systems, including Stanford University School of Medicine, Imperial College London, and The Francis Crick Institute.
Enterprise versions are in private preview with companies including BASF, which is optimizing supply chains, and Klarna, which is developing machine learning models. The U.S. Department of Energy's Genesis Mission and pharmaceutical companies Daiichi Sankyo and Bayer Crop Science are also testing the tools.
Google's trusted tester community includes PhD students, industry researchers, and Nobel laureates. The company has created pilots with conferences including ICML and NeurIPS for automated peer review and scientific validation.
How to access Gemini for Science
Researchers can register interest through Google Labs for gradual access to the three experimental tools. Science Skills is available immediately through Google Antigravity.
Research papers on two of the underlying systems, Empirical Research Assistance and Co-Scientist, were published in Nature.
For researchers looking to build expertise in AI applications across scientific domains, AI for Science & Research courses offer structured learning paths covering research automation and data modeling.
Your membership also unlocks:
