ISTOCK
Virtual scientists in a virtual lab at Stanford University are coming up with unorthodox ways to address clinical challenges, researchers reported on Tuesday in Nature.
The virtual lab is modeled after a well-established Stanford School of Medicine research group, complete with a principal investigator and seasoned scientists, the report says.
As in human-run research labs, the virtual lab has regular meetings during which agents generate ideas and engage in a conversational back-and-forth. They also have one-on-one meetings, allowing the virtual lab members to meet with the virtual principal investigator individually to discuss ideas.
Unlike human meetings, the virtual gatherings take a few seconds or minutes.
When humans tasked the virtual team with devising a better vaccine for the SARS-CoV-2 virus that causes COVID-19, they equipped the virtual scientists with tools and software to stimulate creative “thinking” skills. The virtual scientists even created their own wish list.
“They would ask for access to certain tools, and we’d build it into the model to let them use it,” study leader James Zou said in a statement.
Instead of opting for the usual vaccine design using an antibody, the AI team came up with using a nanobody, an antibody fragment that’s smaller and simpler.
“From the beginning of their meetings, the AI scientists decided that nanobodies would be a more promising strategy,” Zou said.
“They said nanobodies are typically much smaller than antibodies, so that makes the machine learning scientist's job much easier," Zou said, "because when you computationally model proteins, working with smaller molecules means you can have more confidence in modeling and designing them.”
When humans created the AI researchers’ nanobody in a real-world lab, they found it was stable and could attach itself to one of the COVID virus variants more tightly than existing antibodies - a key factor in determining vaccine effectiveness.
Aside from the initial prompt, the main guideline consistently given to the AI lab members was budget-related. Zou estimates that he or his lab members intervene about 1% of the time.
“I don’t want to tell the AI scientists exactly how they should do their work. That really limits their creativity,” Zou said. “I want them to come up with new solutions and ideas that are beyond what I would think about.”