335

Source (Bluesky)

you are viewing a single comment's thread
view the rest of the comments
[-] LwL@lemmy.world 1 points 9 hours ago

It is more appropriate for llms, but not for diffusion models (imagegen). Those are more throw shit at a wall and refine it a thousand times (whereas llms just grab shit that looks similar to what they want). It's why generated images usually look normal at a glance and fall apart the moment you pay attention to details, because the AI judges the whole image to be close enough to training images that match the prompt instead of having any intent behind individual parts.

this post was submitted on 18 May 2025
335 points (100.0% liked)

Fuck AI

2801 readers
1181 users here now

"We did it, Patrick! We made a technological breakthrough!"

A place for all those who loathe AI to discuss things, post articles, and ridicule the AI hype. Proud supporter of working people. And proud booer of SXSW 2024.

founded 1 year ago
MODERATORS