"AI steals art! It's evil, needs to die! I hate it I hate it I hate it waaaaa waaaaaa crycrycry!"
So what do we call someone seeing one person's art, getting inspired by it, and using that as a template or starting point for their own work? Are we going to call that theft? No? Then sit down and shut up.
Are there edge cases in which bad models produced something that was waaaaaay too close to something someone else created or a copy? Absolutely. But there are re-creationists and whatnot in human space too so what's your point?
At the end of the day, show me something that is truly one-hundred percent original, full-stop, no inspiration, no borrowing ideas from someone else's work, techniques or whatever and I'll shut up, but not a single thing in the world that was created artistically didn't get influenced, even in the most basic way, by something or someone else's work.
Besides, if people are using AI to rip off someone else, then the fault isn't the AI, it's the thoughtless dumbass who's using the tech.
And as it's been said a thousand times and will be said a thousand, thousand more times. Bad data will create bad data.
A model that isn't trained right or has very limited sourse material will always reproduce closest to it's training datasets. So enough of the complaining.
I'll give one caveat though as my consession to this debate. Artists should have the right to have their works pulled from datasets as what they make should be their property. Granted the argument that no one ever told someone not to get inspired by the art they see, but this is a much easier thing to do than trying to list in some long as hell credits doc somewhere your name in a list of potentially thousands or tens of thousands of other names, then having to keep track of reuse every model update and re-release.
Seriously peeps, just do some learning, that's all we're asking of you.
Kaylie Fox /
npub1cx…6p284
2023-12-15 19:57:50
Author Public Key
npub1cxzcuk7tnd663pkkcrx2vd3wftuer749tw5t8gvq084dxyyvt74qa6p284Seen on
wss://relay.nostr.bandPublished at
2023-12-15 19:57:50Event JSON
{
"id": "4e669c39721229a87399e739bb1e984ca7bc47ffba682bcb20d94b43a9f5f17d",
"pubkey": "c1858e5bcb9b75a886d6c0cca6362e4af991faa55ba8b3a18079ead3108c5faa",
"created_at": 1702670270,
"kind": 1,
"tags": [
[
"proxy",
"https://tweesecake.social/users/LavenderPawprints/statuses/111586198842666305",
"activitypub"
]
],
"content": "\"AI steals art! It's evil, needs to die! I hate it I hate it I hate it waaaaa waaaaaa crycrycry!\"\n\nSo what do we call someone seeing one person's art, getting inspired by it, and using that as a template or starting point for their own work? Are we going to call that theft? No? Then sit down and shut up.\n\nAre there edge cases in which bad models produced something that was waaaaaay too close to something someone else created or a copy? Absolutely. But there are re-creationists and whatnot in human space too so what's your point?\n\nAt the end of the day, show me something that is truly one-hundred percent original, full-stop, no inspiration, no borrowing ideas from someone else's work, techniques or whatever and I'll shut up, but not a single thing in the world that was created artistically didn't get influenced, even in the most basic way, by something or someone else's work.\n\nBesides, if people are using AI to rip off someone else, then the fault isn't the AI, it's the thoughtless dumbass who's using the tech.\n\nAnd as it's been said a thousand times and will be said a thousand, thousand more times. Bad data will create bad data.\n\nA model that isn't trained right or has very limited sourse material will always reproduce closest to it's training datasets. So enough of the complaining.\n\nI'll give one caveat though as my consession to this debate. Artists should have the right to have their works pulled from datasets as what they make should be their property. Granted the argument that no one ever told someone not to get inspired by the art they see, but this is a much easier thing to do than trying to list in some long as hell credits doc somewhere your name in a list of potentially thousands or tens of thousands of other names, then having to keep track of reuse every model update and re-release.\n\nSeriously peeps, just do some learning, that's all we're asking of you.",
"sig": "d0953cd9534f317d15dd93b636ffbb236089ae07646808ac6eb04bc39f9b0553b59543bc463323ea5c1c8fbaf367fd308ef459243deaa4506b44e1bd5b025665"
}