Google Research announced a suite of AI-powered scientific tools at I/O 2026 that are already helping researchers solve previously unsolved mathematical problems, predict hospital admissions, and accelerate discoveries across neuroscience and cosmology. The toolkit, called Gemini for Science, combines multiple AI systems designed to work alongside human researchers rather than replace them, marking a shift in how artificial intelligence supports the scientific method itself.

What AI Tools Are Inside Gemini for Science?

Gemini for Science is a collection of experimental tools developed in close collaboration with teams from Google Cloud, Google DeepMind, and Google Labs. Each tool addresses a different stage of scientific research, from generating hypotheses to validating findings.

• Empirical Research Assistance (ERA): A research coding system that helps scientists write expert-level software by proposing new concepts, writing code, and evaluating results. ERA searches through thousands of code variants using tree search to optimize performance, accelerating discoveries from neuroscience to cosmology.
• Co-Scientist: A multi-agent system based on Gemini that works as a collaborative AI partner. It uses a coalition of specialized agents to iteratively generate, evaluate, and refine hypotheses. Researchers have used Co-Scientist to tackle antimicrobial resistance, plant immunity, and liver fibrosis.
• Computational Discovery: An agentic research engine built with ERA and AlphaEvolve that generates and scores thousands of code variations in parallel, enabling scientists to rapidly test multiple hypotheses and novel modeling approaches that would take months to explore manually.
• Hypothesis Generation: Built using Co-Scientist, this tool helps scientists synthesize findings across millions of published papers by running a multi-agent "idea tournament" to generate, debate, and evaluate hypotheses with clickable citations for scientific rigor.
• Literature Insights: Built with NotebookLM, this tool helps synthesize findings across scientific literature and structure results for easier analysis.

What Specific Breakthroughs Has This Approach Already Achieved?

The impact of these tools extends beyond theoretical improvements. Empirical Research Assistance has already produced concrete results in real-world applications. ERA helped researchers predict hospital admissions for respiratory illnesses and forecast seasonal runoff across California's river basins, both critical for public health and resource management.

Google also announced that Gemini Deep Think, an advanced agentic reasoning system, recently solved expert-level open research problems in collaboration with mathematicians, physicists, and computer scientists. These breakthroughs included previously unsolved deadlocks in network puzzles, settling a decade-old optimization conjecture, explaining machine learning optimization anomalies, upgrading economic theory for auctions, and resolving physics singularities in cosmic strings.

"In the hands of scientists and researchers, these new types of AI based technologies could change how research is done and catalyze a new era of discoveries," stated Yossi Matias, Vice President and General Manager of Google Research.

Yossi Matias, Vice President and General Manager, Google Research

How Can Researchers Access These Tools?

Google is gradually opening access to Gemini for Science tools and partnering with the global scientific community to ensure responsible advancement. The company is also piloting tools for agentic peer review and scientific validation. Leading scientific conferences including ICML, STOC, and NeurIPS are exploring Google's Paper Assistant Tool (PAT), which reviewed over 10,000 papers in an experimental capacity, helping authors identify critical theoretical gaps or run entirely new experiments based on AI feedback.

• Registration Process: Scientists interested in accessing Gemini for Science tools can register their interest through labs.google/science to be notified as access expands.
• Science Skills Integration: Researchers can perform complex workflows like structural bioinformatics and genomic analyses in minutes rather than hours through Science Skills, a collection of agent capabilities available on platforms like Google Antigravity.
• Peer Review Support: Academic conferences are already testing Paper Assistant Tool to help reviewers identify gaps in submitted research and suggest new experiments.

Why Does This Matter for the Future of Scientific Discovery?

The launch of Gemini for Science represents a fundamental shift in how artificial intelligence supports human research. Rather than automating scientific discovery, these tools are designed to amplify human ingenuity by handling computationally intensive tasks that would otherwise consume months of manual effort. This approach acknowledges that hypothesis generation, literature synthesis, and experimental validation are core to modern research.

Google emphasized that these tools work within what the company calls the "magic cycle of research," moving from hypothesis generation through computational experimentation to validation. This cycle reflects the scientific method itself, suggesting that AI is being positioned as infrastructure for accelerating research rather than replacing the human judgment that drives scientific progress. The emphasis on collaboration with researchers, peer review integration, and gradual access suggests Google is taking a measured approach to deploying these powerful tools in research environments where accuracy and reproducibility are paramount.

Google