Google $GOOGL announced Gemini for Science on Tuesday at its annual I/O developer conference, a collection of AI tools designed to help researchers generate ideas, run computational experiments, and navigate scientific literature more quickly.
The Gemini for Science suite includes three experimental tools covering hypothesis generation, computational testing, and literature review
NurPhoto / Getty Images
Google $GOOGL announced Gemini for Science on Tuesday at its annual I/O developer conference, a collection of AI tools designed to help researchers generate ideas, run computational experiments, and navigate scientific literature more quickly.
The suite offers three experimental prototypes through Google Labs. Hypothesis Generation, created with Co-Scientist, uses a multi-agent system to create, discuss, and assess research ideas, with results backed by clickable citations, according to Google. Computational Discovery, developed with AlphaEvolve and ERA, automates large-scale testing by generating and scoring thousands of code versions at once, covering areas like solar forecasting and epidemiology. Literature Insights, made with Google NotebookLM, allows researchers to search published studies and create structured outputs such as reports, slide decks, and audio or video summaries.
Researchers can register for access through Google Labs, while enterprise organizations have a dedicated route through Google Cloud. Access to all three tools is available via a registration form at labs.google/science; organizations operating at enterprise scale can reach them through Google Cloud. Companies including BASF and Klarna are using AlphaEvolve in private preview, while organizations including Daiichi Sankyo, Bayer Crop Science, and U.S. National Labs are using Co-Scientist to work on research challenges, Google said. Research papers on ERA and Co-Scientist were published Tuesday in Nature.
Google is also launching Science Skills, Digital Trends reported, a bundle that draws on more than 30 life science databases and tools including UniProt, AlphaFold Database, AlphaGenome API, and InterPro. The bundle is available through an agentic platform Google calls Antigravity, and Google said internal testing showed it helped researchers complete a complex analysis of a rare genetic disease caused by mutations in the AK2 gene in minutes rather than hours.
The company said it is working with more than 100 institutions to validate the tools, including Stanford University on liver fibrosis and Imperial College London on antimicrobial resistance. Google has also built a tester community ranging from PhD students to Nobel laureates to stress test the systems, and is piloting tools for agentic peer review with scientific conferences including ICML, STOC, and NeurIPS.
Google also announced Gemini Spark at the same event, a personal AI agent that operates continuously to complete tasks on behalf of users. Spark integrates with Gmail, Docs, and other Google Workspace apps and is described by Google as a personal agent that acts on users' behalf. Google AI Ultra subscribers in the U.S. are set to receive Spark access next week, with expansion to third-party tools via MCP planned for later in the summer. Google reported first-quarter 2026 revenue of $109.9 billion, with Google Cloud generating $20.03 billion for the quarter.
Join 500,000+ readers who start their day with Quartz.
By subscribing, you agree to our Terms of Service and Privacy Policy.