X debate: models may outpredict human experimenters, lowering the bar for virus engineering

cis_female · x · 2026-09-14

In a biosecurity-focused X thread, the author argues that a purpose-built virus research lab would be expensive and slow to build but is almost guaranteed to be created given its value, and that many viruses could be tested in parallel on cell cultures. She adds that models will likely become much better than humans at predicting experimental outcomes, reducing the number of physical experiments needed — which also implies a lower barrier for engineering dangerous pathogens.

Related event: AI Biosecurity Debate: Models May Outperform Humans in Predicting Experiments(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →