🔬 AI Developed a New Approach to Treating COVID
Researchers from Stanford and Chan Zuckerberg Biohub have created a virtual laboratory where AI scientists tackle real medical challenges. In just a few days, the agents designed new nanobody variants (tiny, more efficient antibody fragments) against COVID-19.
👨🔬 How Does This Work?
A human scientist gives a task to the chief AI researcher, who then selects which experts are required to complete it. The SARS-CoV-2 project featured agents with expertise in immunology, computational biology, and machine learning, as well as an AI critic who provided constructive feedback.
The AI leader coordinates the bots' activities and organizes team meetings, which last minutes instead of hours.
"By the time I've had my morning coffee, they've already had hundreds of research discussions," says James Zou, associate professor of biomedical data science, who led a study.
The AI agents code independently, use complex tools such as AlphaFold and Rosetta, and document the results. Humans wrote barely over 1% of the project report's 122,000 words.
🤩 Will AI Help Treat COVID?
In the end, the AI scientist team created 92 nanobody variations, two of which successfully bound to novel SARS-CoV-2 variants in laboratory tests.
The nanobodies bind specifically to the target virus's spike protein and can recognize the original Wuhan strain. Scientists hope that these AI breakthroughs will lay the groundwork for a universal COVID treatment.
