yeah I'm deeply pessimistic about a world where at the very least you don't get good widespread open weight models (and ideally something you can run locally)

also Melty Garfunkel and Oates is going to haunt me

relatedly, I've wondered for a while how many books could be reconstructed out of all the short quotes posted online + analysis to reconstruct ordering.

> He also recently wrote a story about a man with a tooth implanted in his eye. Does that seem probable?

ironically this is actually a real thing that happens

I don't think this is at all the case but the mental image of Trump looking at the Iraq protests and going "a war for oil... sounds like a great idea!" is really funny to me

this means that e.g. i can't ask anywhere "can i get a coding agent that i can run locally and will also act like a doll"

the author of that pr is the jujutsu author, who I assume is trying to prove a point

anyway, giving a serious answer to your original question: "our society has evolved beyond the need for war/money/etc" is a common sci-fi trope, it just doesn't get described as "superhuman". i wouldn't describe e.g. the Culture as "superhuman" even though it's massively better than today

like if someone said "we're working on artificial intelligence when we should be working on artificial empathy" that's not really implying they have any kind of coherent idea for what that *means*, yknow

I don't think quoted OP is saying anything nearly as specific as that so much as, like, taking a phrase and making a substitution without any particular idea of what it would mean

also humans tend to have pesky concerns like "being a moral patient" and "not wanting to exist solely for a single purpose"

bad tsar worse boyars

yeah I was definitely measuring by area taken up on storefront shelves (roughly equivalent to SKUs)

& also if I don't understand the language, I can't evaluate "will this be maintainable in 6 months", which you can't check via tests. this is fine for throwaway code, but the boundary often blurs

(also wonder if LLMs affect learning to evaluate this, since they disconnect you from the pain.)

one of the conceptual problems I have is that if you ask me why I wrote code this way, I can generally remember what I was thinking. if you ask a model why it wrote code that way, it doesn't "remember" its state unless that gets written into the context and checked-in or documented

guy who hates Stalin because he allied with fascists (Churchill and Roosevelt)

not sure about *most* things; there's a lot of homeopathic bullshit, and a bunch of generic "promotes immune system" stuff. it's definitely down to the supplement vs drug regulations, though

yeah, which is why it's funny when I see it on something like Lactaid that actually just works