Beyond Real or Fake: Human–AI Collaboration for Media Verification | NYCU ELITE EP43

Beyond Real or Fake: NYCU Explores Human–AI Collaboration for Media Verification (2026 October EP43) As AI-generated images and videos become increasingly convincing, a simple “real or fake” score is no longer enough. Fact-checkers need to understand what a model has flagged, why it appears suspicious, and whether the evidence supports its assessment. In this interview, Associate Professor Chih-Chung Hsu of the College of Artificial Intelligence at National Yang Ming Chiao Tung University (NYCU) shares the motivation behind his research into manipulated media. His team demonstrates a platform that combines multiple detection models, highlights suspicious regions, and provides explanatory reports to help users examine the results. A real photograph mistakenly flagged because of a face on a sign illustrates why human review matters. Hsu also explores how human observations could help AI recognize unfamiliar forms of manipulation, while the team explains how ONNX and model optimization can reduce computational demands. Looking ahead, media verification will require more than spotting visual flaws. Checking locations, sources, and context—and combining those findings with AI analysis—can provide a stronger basis for judgment. AI detection results provide supporting clues, not definitive proof. Human feedback remains an area of research, and model outputs should be assessed alongside other evidence. | CHAPTERS | 00:00 Introduction: Inside the Research Lab 00:23 The Fabricated Advertisement That Sparked the Research 01:06 Platform Demo: Multiple Models and Visual Clues 01:39 Why Fact-Checkers Need More Than a Score 02:27 When AI Gets It Wrong: Why Explanations Matter 03:33 How Human Observations Could Help AI 04:20 ONNX and More Efficient Model Deployment 04:49 Beyond Visual Flaws: Checking the Real World | STUDENT CONTRIBUTORS | Listed alphabetically by surname, not in order of appearance. You-Cheng Chao Bang-Kang Chen Wei-Chih Huang Chih-Yu Jian Chia-Yu Lin I-An Tsai | LEARN MORE | NYCU ELITE: https://elite.nycu.edu.tw/ NYCU NEWS: https://www.nycu.edu.tw/nycu/en/index Subscribe for more research stories and innovations from NYCU.