back

by sensanaty·3y ago·view on hn ↗
I find these frequently brought-up comparisons of robots to humans pretty funny, personally. I think we can and absolutely should delineate between a human drawing inspiration from their peers and an AI taking a trillion images (far more than any human could possibly ever draw inspiration from) at once, shuffling them all together and spitting out the resulting amalgamation. I see no reason why we should be treating this as if the AI should be treated the same way we treat humans - we definitely shouldn't.

As an example of someone with negative artistic talent of any kind, even if I were to take any one of their works and tried to trace them, it'd come out looking a mess, especially once I came to the portion that required any actual artistic skill like coloring and shading.

With an AI, I save a few of their works, stick it in some blackbox A"I" that gives me 8 trillion different derived images that are in truth just kind of spitting out whatever makes statistical sense to stick together and boom, I've just managed to take a dozen artists' works and gain something out of it.

2 comments
You are drastically underselling the amount of emergent intelligence that comes from $500,000 of Stable Diffusion training.

This is not merely "averaging" input images. It can literally understand human speech.

Maybe I'm being too pedantic, but I'd feel it's far more accurate to say it can interpret human speech within a specific context. There's no understanding going on.
Computers do it more efficiently, yes, but you're not giving an argument as to why being more efficient means there should be different rules.