It's not so much that I disagree with any of this so much as I totally disagree about the timescale. We're iterating on LLMs really fast, which makes it seem li

It's not so much that I disagree with any of this so much as I totally disagree about the timescale. We're iterating on LLMs really fast, which makes it seem like AI as a whole is progressing at warp speed, but there's no indication that that pace is going to translate to the kinds of breakthroughs that would actually make AI an existential threat rather than a new and weird annoyance. As far as GPT-4 already being a potential nuisance in terms of potential social disruption -- yes, sure, but it also has obvious potential benefits, and we seem resolved to total indifference about the way technology is shaping our society (viz. social media). I don't view LLMs as a step change in that regard.