Unable to load image

Training AI on AI generated output leads to model collapse

https://news.ycombinator.com/item?id=41058194

What this implies is future models will be even better at sounding smart but even more likely to hallucinate and give you wrong answers.

The future is r-slurred. :marseywholesome:

62
Jump in the discussion.

No email address required.

Top comment is right:

The key word there is "indiscriminate". All of the big AI labs have been training on synthetic data for at least a year at this point, but they're doing so deliberately.

I don't think the "model collapse" problem is particularly important these days. The people training models seem to have that well under control.

Jump in the discussion.

No email address required.

no they dont

Jump in the discussion.

No email address required.

basically what I as going to say

there's always a certain amount of filtering

Jump in the discussion.

No email address required.

Link copied to clipboard
Action successful!
Error, please refresh the page and try again.