Post #3667980
2026-07-08 08:19 UTC
@bloor@bloor.tw perhaps less so but a lot of the open models are still pretty tainted by the training process. At least you’re not helping OpenAI usage numbers then. I don’t know for sure if any of those are also contributing to the ongoing scraping nightmare but the whole field is a dodgy mess and I don’t want anything to do with it.
Replies (1)
-
@bloor@bloor.tw 2026-07-08 08:23
@bencc@morehammer.uk So it is the imbibing of massive quantities of source material, without consent, that is the problem. OK. That does make sense. But let's imagine I asked an artist friend to make me a silly meme for my toot. I'd say "oh I want it to look like count binface in front of clacton pier" and she'd go straight to google images, view images of binface, and Clacton, without the originator's consent, then make something by hand. I'm wondering what makes that different and more acceptable? Scale?