> We must move past vague discussions of “AI” and instead isolate the specific types of systems that are creating problems.
This is absolutely correct and its surprising how rarely you see commenters in the media push to go into the specifics on this. AI is just matrix math and its what you connect that math to that matters, we should talk a little more about what we are willing to connect AI to and little less about how scary or capable the models appear.
A good start might be removing virology info above the undergraduate level from training sets. The main "kills everybody" threat involves biological viruses.
AI-driven hacking is a problem, but it's really just nation-state level hacking available to smaller companies or wealthy individuals. Everything needs to be toughened up to the point that doesn't work. Ask countries that have been up against Russian hacking for years, such as Estonia.
Beyond that, most of the threats are not that serious. Below the level of organized crime or corrupt government.
There's no evidence that a doomsday virus can be created. There's a tradeoff between how infectious and how deadly a disease is. You would have to develop a two-stage virus, and that would require a large lab or labs with material being shipped there. Something humans would definitely notice. Even then, how would the virus reach literally everyone? People will take measures to not become infected once the threat has been identified, and some people and groups are isolated as it is. And no virus kills everyone. Rabies comes closest at 99%. But unlike what the zombie genre would have you believe, it has a poor means of delivery.
This isn't like vibe coding. The AI would have to make the virus in laboratory conditions and then disseminate it globally without prior detection.
You don’t need to attack humans directly. Simply making and distributing a virus or fungus that successfully damages the grasses that we use for the bulk of our calories would do the job.
You'd need to hit wheat, corn AND rice at the same time. And we probably have 10 others we're not using because those 3 are already enough. It would be a tragedy but we can eat lots of things.
Yes, with AI I will be able to simply build a doomsday machine. Claude will just tell me how to build a supply chain, circumvent regulations and do this without anyone finding out.
jimmyjazz14 · · focus · HN ↗
This is absolutely correct and its surprising how rarely you see commenters in the media push to go into the specifics on this. AI is just matrix math and its what you connect that math to that matters, we should talk a little more about what we are willing to connect AI to and little less about how scary or capable the models appear.
Animats · · focus · HN ↗
A good start might be removing virology info above the undergraduate level from training sets. The main "kills everybody" threat involves biological viruses.
AI-driven hacking is a problem, but it's really just nation-state level hacking available to smaller companies or wealthy individuals. Everything needs to be toughened up to the point that doesn't work. Ask countries that have been up against Russian hacking for years, such as Estonia.
Beyond that, most of the threats are not that serious. Below the level of organized crime or corrupt government.
goatlover · · focus · HN ↗
This isn't like vibe coding. The AI would have to make the virus in laboratory conditions and then disseminate it globally without prior detection.
foobiekr · · focus · HN ↗
oblio · · focus · HN ↗
foobiekr · · focus · HN ↗
m4x · · focus · HN ↗
plastic-enjoyer · · focus · HN ↗