At this year's Google I/O 2026 conference, Google officially launched a set of experimental AI tools called Gemini for Science, designed to help researchers optimize the complete scientific research process from formulating hypotheses to reviewing literature. This suite is positioned as a new attempt to use generative and inferential artificial intelligence to improve the efficiency of scientific discovery. It has begun to gradually open application channels to scientific research institutions and corporate users.

Gemini for Science currently includes three core functions: Hypothesis Generation, Computational Discovery, and Literature Insights. The hypothesis generation tool assists researchers in conceiving new theoretical assumptions or asking challenging questions in the early stages of scientific research by retrieving and analyzing millions of scientific papers. The generated conclusions will be accompanied by clickable references for rigorous verification. Google claims that these conclusions are "deeply verified and equipped with clickable references to ensure absolute rigor."

After forming a scientific research hypothesis, researchers can use computational discovery tools to verify the hypothesis on a large scale. Google describes this feature as "a search engine with agent capabilities" that can automatically generate thousands of test and experimental plans, greatly reducing the work time that originally required manual design and screening, thereby speeding up the iteration cycle of theoretical screening and experimental design.

The literature insight function focuses on the literature review and knowledge acquisition process in the scientific research process. By automatically sorting out a large amount of scientific literature, it can generate easier-to-understand reports and summaries for researchers. In addition to text reports, this tool can also output overview content in the form of infographics, audio or video, helping researchers more quickly grasp cutting-edge progress and relevant background knowledge, and lowering the threshold for reading complex scientific research materials.

As part of the Gemini for Science program, Google also launched Science Skills tools to integrate professional databases and tool resources in the field of life sciences. According to the official introduction, Science Skills can mine insights from more than 30 major life science databases and tools, shortening the originally complex and highly manual scientific research workflow to "complete in minutes" instead of the hours or even longer it took in the past.

In terms of openness, Google stated that it has gradually opened up the early trial qualifications of Gemini for Science to the outside world from now on. Interested researchers can fill out a form on the Google Labs website to apply for access. In addition, Google also plans to provide these AI scientific research tools to enterprise-level organizations through Google Cloud, with a view to promoting the application of AI in scientific research and industrial R&D on a larger scale.