AI Tool Helps Scientists Crack Cancer's Toughest Problem

Scientists just got a powerful new research partner that never sleeps, and it's already suggesting solutions to problems that have stumped experts for years.

Anna Pertl, a biochemist at the Whitehead Institute in Massachusetts, typed a question into an AI system called Co-Scientist before heading home for the night. She wanted fresh ideas for attacking MYC, a protein that fuels most cancers and has resisted every treatment attempt.

By morning, the AI had reviewed more than 700 scientific papers and proposed something remarkable. Instead of trying to dissolve the molecular clusters that activate the cancer gene (the approach most researchers pursue), it suggested gluing them together into a gooey mass that would shut down the gene entirely.

"We had certainly never thought about anything like this," says Kalon Overholt, a bioengineer who worked with Pertl on the project. The idea flipped their entire approach on its head.

Co-Scientist works differently from typical chatbots. It launches multiple AI agents that search papers, evaluate competing theories, refine hypotheses, and test ideas against published evidence. The process requires massive computing power and can take hours or even days.

Research teams are already putting these AI collaborators to work. Scientists have used Co-Scientist to discover drug combinations that kill leukemia cells and identify treatments that regenerate diseased liver tissue in the lab.

Why This Inspires

The real excitement isn't about replacing scientists. It's about freeing them from the tedious work of combing through mountains of research so they can focus on the questions that matter most.

"We imagine it to be like a collaborator, a partner with you," says Vivek Natarajan, an AI researcher at Google who helped develop Co-Scientist. Le Cong, a molecular geneticist at Stanford and co-founder of similar AI company Phylo, puts it simply: "We're trying to move humans up the value chain."

Several organizations are now developing these research assistants. Companies like Anthropic, OpenAI, FutureHouse, and Phylo are rolling out systems that can tackle tasks once reserved for human scientists alone.

The shift could change what makes a scientist valuable. For generations, scientific progress depended on people who could pose difficult questions, devise experiments, and interpret results. As machines handle more synthesis and analysis, human scientific judgment becomes the scarce resource: knowing which questions are worth asking and which ideas are worth pursuing.

The technology does create new challenges. Scientists will need deep expertise to evaluate AI-generated ideas, even as students get fewer chances to build that knowledge by doing hands-on research work. But the promise of accelerating breakthroughs while liberating researchers from routine tasks makes this a transformation worth watching.

For Pertl, who jokes she could complete a triathlon in the time it takes Co-Scientist to work, the partnership is already proving its worth.