wake up, see another really bad AI art take on Twitter

why does this keep happening

"*looking at a waterfall* wow, nature is so beautiful!"

"um, ACTUALLY, only humans are capable of generating beauty. nature isn't conscious and so can only produce shallow imitations of real beauty"

"generative AIs just combine images" is a really common misconception which would be absolutely *wild* if true. like, fitting a billion-image training set in a 5GB model? that's just not possible

"ash why are you so consistently mad about this"

because I used to work on an ML system and it's something I know a lot about and it's fucking frustrating seeing people give these horrifically incorrect explanations!

and what's worse is when people react to people saying "no, that's not how it works" with "oh so you want artists to starve? where's your nft collection techbro LMAO" because online has poisoned people's brains and turned engagement with actual real people into Sick Burns

and like people don't make shit up about how, say, GPUs or BGP or CPU microarchitecture works, but as soon as it's about AI/ML, everyone's suddenly a fuckin expert about how it can't create new inputs because it doesn't have the magic creativity molecule or some shit

it's like how people react to tech dipshits going "AI is unbiased and will never make mistakes" with "AI is inherently biased, nobody has ever thought about debiasing techniques until me because I'm so much smarter than those stupid techbros. also, we should use humans instead, because humans are well known for having no unconscious bias and being able to explain their decisions"

basically I think at some point it became "cool" to dunk on AI tech and this is the result, because AI suffers from the constant reputation problem of "oh, X? that's not Really AI. if it was Real AI then it'd [constantly-shifting goalposts]"

(also, it leads to people having takes about copyright like "style should be copyrightable" and "you shouldn't be able to train an AI on someone else's content without their permission no matter what" that would, frankly, lead to a *massive* expansion of what copyright would cover in a way I don't think people entirely think through.)


also that "AI art can't be copyrighted'" thing going around is based on a *massive* misread of what happened

what actually happened is that someone tried to register a piece of art with the program he wrote as the sole author. the copyright office said "okay, well, you're claiming the AI as the author, and only humans can hold copyright. so no." the office did not talk about whether it would be copyrightable if he registered it in his own name, and says it was explicitly beyond scope to decide this!

> Because Thaler has not raised this as a basis for registration, the Board does not need to determine under what
circumstances human involvement in the creation of machine-generated works would meet the statutory criteria for
copyright protection.

this is like me trying to register a piece of art with my laptop as the copyright owner, getting denied, and then news articles saying "us copyright office rules digital art can't be copyrighted'"

Sign in to participate in the conversation
inherently digital

a very robotic single-user instance