AI code judges don't know when they're guessing
A new study finds AI judges rating code quality are often guessing, and proposes a way to detect when a verdict has no real basis.
A new study finds AI judges rating code quality are often guessing, and proposes a way to detect when a verdict has no real basis.