Post #2158784

I had been ignoring AI security, assuming it was just a modernized version of traditional systems security. Interesting, sure, but not really suited to my patient, math-oriented mindset. But honestly, the security issues in AI don’t revolve around systems and software as much as I expected. There’s an added layer of “psychology,” which is actually quite intriguing. With increasing complexity of agentic AI, at some point a person with better social understanding will be a better AI security expert. Dealing with all this, I’m starting to see new layers in my arguments against the idea that silicon-based ASI will surpass us at the evolutionary level. At the same time, I’m finding stronger reasons to support the idea of a more integrated species–society.