Google has revealed a new system called “Co-Scientist” that it says could lead to major new breakthroughs.
The artificially intelligent tool is powered by similar technologies to those in chat-based large language models, such as Google’s own Gemini and competitors such as ChatGPT. But it is specifically devoted to helping find new research breakthroughs.
Scientists who have used it describe it as being like having an especially well-read and helpful colleague. It has already been used by researchers at Imperial College London who were able to recreate their own work much more quickly than they would have done without using the system.
To use the system, an expert scientist uses normal language to give it a research goal. The system then checks through the published literature and synthesises it, but can also assess its own findings and propose new hypotheses and possible experiments that might validate them – all while citing literature and explaining its proposals.
Google says that the system is built specifically to be collaborative, rather than automating research entirely through AI. Scientists are able to chat with it, providing feedback and altering its understanding that might improve it over time.
The company undertook a range of ways of evaluating the system, including asking experts to come up with 15 “challenging and open research goals in their field” and have the co-scientist produce possible solutions. Those experts rated the new tool above other existing tools.
Scientists speculated that the system could be a new way of quickly thinking up new hypotheses and new breakthroughs on a range of important topics. It has already been used to address research on antimicrobial resistance, for instance, which the World Health Organisation says is one of the biggest threads to global wellbeing and safety.
“Laboratory science is resource-intensive, and with global challenges like antimicrobial resistance looming, it’s clear we need to do more with less and speed up new discoveries,” said José Penadés, who co-led the experimental work at Imperial’s Department of Infectious Disease.
“When the Google research team approached us to to test its AI platform, we realised we needed to task it with the same scientific questions that we had already explored ourselves and used as the basis of our experimental work.
“This effectively meant that the algorithm was able to look at the available evidence, analyse the possibilities, ask questions, design experiments, and propose the very same hypothesis that we arrived at through years of painstaking scientific research, but in a fraction of the time.
“This type of AI ‘co-scientist’ platform is still at an early stage, but we can already see how it has the potential to supercharge science.”
Professor Penadés suggested that the system was “scary” because it was not possible to know exactly how the system might evolve. While it could not do the job of a scientist immediately, for instance, it might eventually automate some of their work.
Google itself noted in its announcement that the innovation “opens numerous questions”, including how scientists will be able to quickly evaluate its many hypotheses, as well as how to credit the system and the ways it will change practical problems such as who is credited for research and how to apply for funding. It also noted concerns about how to ensure creativity in scientific research, noting that the co-scientist led to questions about “preserving diversity and serendipity in hypothesis generation”.
But scientists said that the system could also provide entirely new ways of meeting urgent questions that researchers are currently rushing to answer.
“The world is facing multiple complex challenges – from pandemics to environmental sustainability and food security,” said Mary Ryan, vice provost for research and enterprise at Imperial. “To address these urgent needs means accelerating traditional R&D processes and artificial intelligence will increasingly support scientific discovery and pioneering developments.”
At the moment, Google is only making the system available through a “trusted tester programme” that is intended to ensure it is used by responsible people. Researchers are able to join that programme through Google.