Google launches Gemini for Science

- Google unveiled Gemini for Science at its I/O 2026 conference on May 19, adding experimental research tools in Google Labs and Science Skills in Antigravity. (blog.google) - Google said the launch centers on three Labs prototypes — Hypothesis Generation, Computational Discovery and Literature Insights — built with Co-Scientist, AlphaEvolve, ERA and NotebookLM. (blog.google) - Researchers can express interest through Google Labs’ science page, while Google’s I/O materials place the tools alongside Antigravity 2.0 and Gemini 3.5 updates. (labs.google)

Google used I/O 2026 to package several of its research-focused AI efforts under a new label, Gemini for Science. In company materials published May 19, Google described it as a collection of science tools and experiments designed to help researchers with core parts of the scientific method. (blog.google) The launch spans two surfaces: new “Science Skills” in Google Antigravity and three experimental tools hosted in Google Labs. Google announced the broader I/O slate from Mountain View, California, as part of its developer conference. ### What exactly did Google launch? Google’s May 19 post said Gemini for Science includes “new Science Skills in Google Antigravity and three new experimental tools on Google Labs.” The company said those tools are meant to “accelerate core steps of the scientific method” and are built with Co-Scientist, AlphaEvolve, Empirical Research Assistance and NotebookLM. (labs.google) Google Labs’ science page lists the three experiments as Hypothesis Generation, Computational Discovery and Literature Insights. The page says users can express interest in the tools through Labs, indicating the products are still in an experimental access phase rather than broad general release. (blog.google) ### What do those tools do for researchers? Hypothesis Generation is described by Google Labs as a multi-agent system that helps identify knowledge gaps and propose testable research plans. Google said in its launch post that the tool simulates the scientific method through an “idea tournament” that generates, debates and evaluates hypotheses, with claims supported by clickable citations. (blog.google) Computational Discovery is positioned as an agentic research engine for computational experiments. Google said the prototype can generate and score thousands of code variations in parallel, while the Labs page says researchers define optimization tasks and evaluation requirements and then use the system to explore modeling directions that would be difficult to test manually. (labs.google) Literature Insights is aimed at paper search and synthesis. Google Labs said it can search scientific literature, structure results into tables, extract metrics and variables, and generate reports, slide decks, mind maps, audio overviews and infographics with citations linked back to source text. (blog.google) ### Where does Antigravity fit into this? Google’s I/O developer materials describe Antigravity as the company’s “agent-first development platform.” In a separate May 19 post, Google said it was expanding that ecosystem with Antigravity 2.0, a desktop application designed to orchestrate multiple agents, plus command-line, SDK and enterprise integrations. (blog.google) The Gemini for Science post ties the science launch directly to that platform by adding “Science Skills” inside Antigravity. Google did not spell out every individual science skill in the excerpts available on its public announcement pages, but it presented Antigravity and Labs as the two main delivery channels for the new science offering. (labs.google) ### How does Google describe the strategy behind it? Pushmeet Kohli, Google Cloud’s chief scientist and a Google DeepMind vice president, and Yossi Matias, vice president at Google and general manager of Google Research, wrote that Google sees scientific discovery being aided by “general agents” rather than only narrow specialized models. (blog.google) They said the aim is to expand “the scale and precision of scientific exploration.” Google’s wording frames the tools as assistants for time-consuming research work rather than replacements for scientists. The launch post says AI can help handle complex tasks so researchers can focus on identifying and pursuing the most important scientific problems. (blog.google) ### What happens next? Google Labs’ science page says interested users can sign up to try the experiments, making Labs the immediate access point for the three prototypes. Google’s I/O materials also place Gemini for Science alongside the company’s May 19 rollout of Gemini 3.5 updates and Antigravity 2.0, which Google said is now part of its developer offering. (labs.google) (blog.google)

Get your own daily briefing

Scout delivers personalized news, insights, and conversations tailored to your role and industry.

Download on the App Store

Shared from Scout - Be the smartest in the room.