Samuel
@samuel.fm
apropos of nothing, openai does not train its models on stuff you send it via the API
After an image has been processed by the model, it is deleted from OpenAl servers and not retained. We do not use data uploaded via the OpenAl API to train our models. 17 13 79
Daniel Kelly
If you had made it "whether or not the scrabble score of your username is odd or even" it would've been fun when people figured it out and started trying to game it.
2
Thorne 🌸 needs a job!
To certain folks, new tech just become an amorphous blob of badness where all criticisms apply at all times in all circumstances. This won't change how they feel any more than stuff like, "this chain uses Proof of Stake, which is very power efficient."
14
Dr. Jimbosaur, Esq.
"We would never stoop so low as to use what you've sent us via the API as training data. We train our algorithms exclusively on artisanally stolen images from organic non-consenting creators."
1 21
Rabbit
i don't even understand how people think this would work. if we're talking training data for ai art then that needs pairs of images and descriptions. you can't just throw a bunch of images at a computer and not tell it what any of them are and expect it to be able to do anything with them
2 5
Rabbit
stirring hundreds of terabytes of raw image data in a cauldron and midjourney 8 pops out apparently
4
Joel Berger
Which is not the only problem with "AI." There are so many ways you could have done this and you managed to pick the one that most of the platform was going to hate.
1 1 7
John ✦°
What I've been finding funny is the people on Insta or Threads complaining about training data when they are on a platform that uses all its data to train their models.
2
ǝuuɐſ
logging on the decentralised discourse network to discourse, ready to discourse, discourse gonna discourse