Post #3503472
2026-06-24 03:09 UTC
When an AI is confident it’s right, I become even more convinced it’s wrong. An AI that isn’t confident will try to verify when challenged, or admit it’s wrong (even if it’s not) but one that doesn’t know how to verify what it’s saying will just blindly insist that it’s totally sure and make up some nonsense to justify it.
Replies (1)
-
@DevDave@piefed.social 2026-06-24 06:39
I find it fascinating how the architecture of LLM’s are locked into “Confidently Just make shit up” once it has gone beyond the limits of its parameters.