False Positives

Concept

A critical problem with AI models, especially in defense, where high false positive rates (like 30% in Mythos) render them unreliable for critical functions without proper mitigation.

Mentioned in 1 video