Google AI Finds New Cancer Drugs, Gives Doctors 40% More Time

What if AI could help cure cancer instead of just taking jobs?

Google's head of research, Yossi Matias, is betting the future of artificial intelligence looks less like automation and more like scientific breakthroughs. He's leading two ambitious projects that are already showing results.

The first system, called Co-Scientist, reads through mountains of scientific papers and generates new research ideas. It then ranks those ideas so scientists know which ones are worth testing. The second, ERA, helps automate the tedious work of building models and running experiments.

The systems are already making discoveries. According to a new paper in Nature, Co-Scientist identified promising drug candidates for acute myeloid leukemia and uncovered new information about antimicrobial resistance.

Matias calls it having "a polymath in your pocket." The goal is giving every scientist, even students, access to a virtual research lab that can sift through endless studies and point them toward the most promising questions.

The stakes are enormous. Scientists often spend years chasing weak hypotheses. If AI can help them focus on the right ideas faster, treatments for cancer and rare diseases could arrive years sooner.

Google already tested similar AI technology with the UK's National Health Service for breast cancer screening. The AI acted as a "second reader" of mammograms and caught 25% of cases doctors missed. It also gave doctors back 40% of their time.

That was five years ago with older technology. Today's AI is far more powerful.

Matias doesn't see AI replacing scientists. Instead, he imagines junior researchers leading teams of AI collaborators the way senior scientists currently manage human research teams. What once took decades of experience could become available to practically every researcher.

Why This Inspires

The narrative around AI has grown increasingly dark, focused on job losses and existential threats. But Matias offers a different vision: technology that multiplies human capability rather than replacing it.

"The power, the effectiveness of every researcher is going to be multiplied," he says. The most exciting part? We're nowhere close to answering all the important questions about human health.

This approach could democratize scientific discovery. Graduate students and researchers at smaller institutions could pursue questions previously reserved for elite labs with massive teams.

The timeline matters too. While Matias stresses that curing cancer will take time, the NHS breast cancer study shows these tools work right now. The challenge isn't the technology anymore. It's figuring out how to integrate these discoveries into healthcare systems worldwide.

Matias remains optimistic: "I believe we are going to get to a state where we can actually identify everything that needs to be identified about diseases and find whatever solution that can be found."

For a technology often framed as humanity's competitor, that sounds like the partnership we've been waiting for.