Why human scientists, not AI, still control the fate of biosecurity
Get Quartz in your inbox
Free daily briefing on global business news.
Why human scientists, not AI, still control the fate of biosecurity
AI models have written the genome of a working virus for the first time, though the ones that came alive attack only harmless lab bacteria
Joan Cros / NurPhoto via Getty Images
Samuel King was watching petri dishes in the early hours of the morning when he started seeing clear spots spreading across a layer of bacteria. A virus his team had designed on a computer was alive and killing. "We were starting to see these clear spots and it was just extremely exciting," King, a PhD student in the Stanford University lab behind the work, told the BBC. When his colleagues saw the results, the room spontaneously burst into applause, lab leader Brian Hie recalled.
King and his colleagues at Stanford and the Arc Institute had just done something no one had done before. They used AI to write the complete genome of a working virus from scratch. Two AI models designed nearly 300 viral genomes, and 16 of them turned into living viruses that the team built and tested, according to an analysis by CAS, a division of the American Chemical Society, of the study the team published in the journal Science in August. "It's something that can replicate and have other functions inside cells," King told the BBC. "This was new territory for us."
The viruses they built can't hurt a person, and that isn't by chance. The team aimed AI at ΦX174, a virus that attacks only a harmless strain of gut bacteria, and chose it precisely because scientists have studied it for decades and long used it as a safe laboratory host.
The history of the tiny long-studied virus
AI didn't dream up viruses out of nothing. It started from one of biology's most familiar targets, a phage whose genome runs only about 5,400 DNA letters long and carries instructions for making just 11 proteins. That makes........
