Gemini for Science

Gemini for Science is Google's Gemini AI model tailored for research work. It's built to help researchers keep up with newly published papers, turn research ideas into working code, and spot new hypotheses while looking through data.

Think of it as a diligent new assistant joining the lab. Each morning it scans and organizes newly posted papers, writes code once a researcher describes the experiment they want to run, and even suggests hypotheses while reviewing data together with the researcher. But since it hasn't been "hired" full-time yet — it's still at the "Labs prototype" stage, like an intern on trial — details like which paper databases it searches or which programming languages it codes in haven't been disclosed.

Google has recently been moving away from having a single version of Gemini handle everything, instead fine-tuning separate versions for different fields. It previously trained a version tailored for medical consultations, and now it has released one for researchers.

The tool appears in coverage alongside Google DeepMind's announcement that "this new Labs prototype streamlines everyday scientific work." Details such as which paper databases it connects to, or its accuracy and processing speed, haven't been disclosed yet, so it shouldn't be mistaken for a finished commercial product. As the term "Labs prototype" suggests, it's still in a testing phase.