Some leaders in artificial intelligence are urging a slowdown for safety, and the Asilomar conference is a reminder that a voluntary pause can work when a field faces shared risks that no single lab or company can shoulder alone.
In the early 1970s, biochemist Janet Mertz was working on recombinant DNA, a technique that allowed scientists to cut DNA from one organism and insert it into another. The promise was enormous, including mass-produced insulin, improved crops and potential treatments for disease.
That promise came with fear. As the story is recounted, Mertz planned an experiment involving a cancer-causing virus and E. coli. When she discussed it with fellow biologist Bob Pollack in 1971, he grew alarmed about the potential for a dangerous outcome, even raising the specter of a superbug that could spread cancer. Mertz’s adviser ultimately canceled the experiment.
The tension was clear. Scientific progress was accelerating, but no single lab wanted to slow down and risk losing discoveries or accolades. In 1974, senior researchers, including Mertz’s supervisor, issued an open letter calling for a pause on the most cutting-edge recombinant DNA work so the community could meet and establish safety standards.
Asilomar conference and AI safety
The resulting gathering, now known as the Asilomar conference, brought leading scientists together to debate how to handle the new technology. Recordings preserved by the National Academies capture those discussions, including comments from bioethicist Alex Capron, who warned that freedom of inquiry did not extend to causing harm and that, absent self-governance, Congress could step in.
Participants ultimately agreed on lab protocols and guardrails that allowed research to continue with clearer boundaries. The key takeaway was that a coordinated pause, combined with shared standards, can resolve a collective action problem when the stakes are high.
For today’s AI leaders in Hawaiʻi and beyond, the Asilomar conference shows that rivals can align on safety without halting innovation entirely. The lesson from the biology lab is that getting competitors in the same room, setting rules and resuming with care is possible when the risks demand it.
More on these lessons appears in a recent Planet Money episode that explores how the Asilomar approach could inform AI governance, echoing concerns raised as big banks flag AI financing risks but keep funding.
2 Responses