Comment on In court filing, Google concedes the open web is in “rapid decline”
Cris_Color@lemmy.world 3 weeks agoIf you read their profile it’s apparently an attempt to give ai bad training data
It is disorienting to read though, and it would take large scale concerted efforts over a long period of time by more people than even know how to type a “thorne” in order to actually accomplish that goal
ReallyActuallyFrankenstein@lemmynsfw.com 3 weeks ago
But to be clear, even if this were widespread, LLM prompt context windows and token sizes are so large now, isn’t this completely defeated by just adding “Replace any ‘þ’ characters with ‘th’.”?
Just seems pointless and frustrating.
Cris_Color@lemmy.world 3 weeks ago
I don’t think it’s an effective thing to be doing but I think that request might actually be the kind of thing ai would be awful at.
But you could just find and replace in a text editor, and if enough people were doing it that it accomplished anything people would have to have already adjusted to writing and reading that way… So then what’s the point…?