flâneur

Egan Peltan (@eganpeltan): "As a scientist I need to pushback on the cursory dismissal of viral creation risk. The arguments against AI-assisted viral risks are weak and largely miss the point. Synthetic viral risk is real now and LLMs can lower the barriers to create novel ePPPs The post makes the follo…"

substack.com · 732 words · saved by 1 readers

As a scientist I need to pushback on the cursory dismissal of viral creation risk. The arguments against AI-assisted viral risks are weak and largely miss the point. Synthetic viral risk is real now and LLMs can lower the barriers to create novel ePPPs The post makes the following claim: “The 1918 flu infected roughly a third of the world’s population and killed an estimated 50 million people. Releasing it today would be very bad. It would not be 1918-bad. Most people’s immune systems recognize flu descendants. We also now have antivirals and antibiotics to treat the secondary bacterial pneumonia that killed most victims. Today’s vaccines offer cross-protection.” The partial protection claim is unsubstantiated. Most evidence points in the opposite direction. I hope this unsubstantiated argument isn’t a paraphrase of expert calls. I know the author isn't a scientist, but the experts - who should be named - should know better. Flu vaccines and flu infection offers little protection against new strains in future years. Partial ancestry isn’t really protective against subsequent flu infections. This is partially why the flu vaccine is remade yearly! The experts should be familiar with the foundational 1918 flu reconstruction work. There is good evidence that modern H1N1 strains offer little-to-no protection against the chimeric 1918 flu constructions. Figure 4 from Kobasa et al. 2024 uses a partial reconstruction of a 1918 virus (HA chimera) to test cross-protection. Solid diamonds are the 1918-derived (HA chimera); open diamonds are a modern H1N1 strain… This suggests there’s little-to-no little cross-protection from modern flu against 1918 flu. We can’t be confident that we can prevent or treat 1918 flu with modern anti-infectives. The seminal 1918 reconstruction paper doesn’t assess neutralization with modern sera (nor should they have - they shouldn’t have made the virus in the first place!). Follow-ons like the viral archaeology of Worobey, Han, and

As a scientist I need to pushback on the cursory dismissal of viral creation risk. The arguments against AI-assisted viral risks are weak and largely miss the point. Synthetic viral risk is real now and LLMs can lower the barriers to create novel ePPPs The post makes the following claim: “The 1918 flu infected roughly a third of the world’s population and killed an estimated 50 million people. Releasing it today would be very bad. It would not be 1918-bad. Most people’s immune systems recognize flu descendants. We also now have antivirals and antibiotics to treat the secondary bacterial…

saved by

related reading