Teaching AI models to say “I’m not sure”
Asset Info
CreatorN/A
Registration TimeLoading...
RegistrarMIT News - Artificial intelligence
Capture TimeLoading...
GeolocationN/A
File TypeJPEG
Source TypedigitalUpload
Details
Abstract
A new training method improves the reliability of AI confidence estimates without sacrificing performance, addressing a root cause of hallucination in reasoning models.
LicenseN/A
Used Bynews.mit.edu...
Mining PreferenceN/A
Integrity Proof