
Can humans be trained to spot AI-generated deepfake images?
Researchers have found that people can be trained to distinguish AI-generated faces from real ones, improving accuracy from 40% to 80% in about an hour. The training focuses on subtle perceptual qualities like symmetry, proportionality, attractiveness, distinctiveness, expressiveness, and memorability, as AI struggles to perfectly replicate human quirks. This skill is crucial given the rise of AI deepfake scams and potential political espionage.







