3 ms·
Could AI-designed proteins be weaponized? Scientists lay out safety guidelines
- gilleain 3y agoWhy provide an image without a reference to what it is? Irritating. Anyway, through some lazy detective work, I think it looks like : https://alphafold.ebi.ac.uk/entry/Q8W3K0 https://alphafold.ebi.ac.uk/entry/Q8W3K0 Which I then think might be a LRR superhelix topology (3.80.10) under the CATH classification: https://www.cathdb.info/browse/sunburst?from_cath_id=3.80.10 https://www.cathdb.info/browse/sunburst?from_cath_id=3.80.10
- gcanyon 3y agoWe are heading toward times that will make the Cold War and Mutual Assured Destruction look like child's play. Building nuclear weapons is still very difficult almost a century after their first development. By contrast, within the next 30 years (20? 10?) malicious bioweapons, AI, nanotech, flawless disinformation, and probably other threats I'm not thinking of are going to be things a high schooler with a grudge can develop, and certainly within the reach of a terror group or a hostile government even sooner. And there won't be the obvious signs there are with nuclear -- high speed centrifuges, fissile material, etc. We are not prepared. At. All. If we survive, a paradise of technological advancement awaits. But that's a big "if."