AI models spot deepfake images, but people catch fake videos
SMRTR summary
Researchers tested humans versus AI algorithms in detecting deepfakes, finding AI excelled at spotting fake images with 97% accuracy compared to humans' 50% chance-level performance. However, humans surprisingly outperformed AI when identifying fake videos, achieving 63% accuracy while algorithms dropped to chance levels, suggesting future deepfake detection requires human-AI collaboration.
SMRTR provides this summary for quick context. The original article belongs to Science News.
Read the original article