4 ways researchers are collaborating with Co-Scientist to solve big problems

Learn more about Co-Scientist and the impact it’s already making.

Your browser does not support the audio element.

This content is generated by Google AI. Generative AI is experimental

We recently published our latest research on Co-Scientist, a collaborative AI designed for structured scientific thinking to help researchers develop new hypotheses in life sciences and beyond.

The system is made up of a coalition of specialized agents that work together in three distinct phases. First, Co-Scientist generates ideas through agents that propose hypotheses and explore a wide variety of research avenues. Next, it debates ideas, with one agent that acts as a virtual peer reviewer before another pits vetted ideas against each other in an “idea tournament.” Finally, it evolves ideas, with agents that refine, combine and improve the best hypotheses, as well as agents that synthesize the research for a human scientist. A supervisor agent ties the system together, breaking down high-level research goals into individual tasks, allocating resources and coordinating specialized agents to work in parallel.

Since sharing early research last year, we’ve been developing and testing the system with scientists and research teams globally tackling complex scientific problems. Read about just a few ways it’s already having an impact:

Co-Scientist will be available to researchers through Hypothesis Generation, a new experimental tool developed across Google DeepMind, Google Research, Google Cloud and Google Labs. Learn more about Co-Scientist and how researchers are using it on the Google DeepMind blog.