Hacker News Reader

Comment by autoexec | Hacker News Reader

> What's next? Shutting down an AI is murder?

I can see a possible future where AI actually exists and shutting one down could be murder. At that point it would even be a good thing to grant the AI personhood. What passes for "AI" these days doesn't come anywhere close to that, but I wouldn't say it could never happen.

View on HN

About 95% of the conversation about “AI” has this problem right now: there are some interesting theoretical legal and social implications from AI, but what we have right now are LLMs, not AI. They can’t replace your workers, they can’t make art, they can’t hold copyright, not because the law doesn’t treat them as people, but because they’re a fancy autocomplete algorithm that spits out text convincing enough to spike the pareidolia tendency that’s led to humans assigning agency to every other inanimate object that’s ever sparked an emotional reaction in us too.

With image AI there are structural editing tools that can include the creator drawing guide images. I think there's a point at which this makes things a bit more like collage of found media, and I believe this is more of an authored creation when compared to Richard Prince's using someone else's Instagram selfie and repurposing that as his art by adding a comment to it and screen-grabbing it. What is and isn't art is sometimes to me, clearer than copyright ownership and sometimes vice-versa.

https://www.theverge.com/2015/5/30/8691257/richard-prince-in...

> what we have right now are LLMs, not AI

LLMs are a subset of AI. The field has been called AI for over 70 years.

Let's try to at least keep the terminology correct on HN.

Reply on news.ycombinator.com