Perceived authenticity drives gaze behavior when watching AI-generated videos of physical scenes
Danqing Shi, Lan Jiang, Katherine M. Collins, Shangzhe Wu, Ayush Tewari, Miri Zilka · 2026
Read Perceived authenticity drives gaze behavior when watching AI-generated videos of physical scenes by Danqing Shi, Lan Jiang, Katherine M. Collins, Shangzhe Wu, Ayush Tewari, Miri Zilka (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13482358.txt.
Description
Shi and colleagues tracked the eyes of 40 viewers in 2026 to learn how people watch AI-generated videos of physical scenes, and what gives the fakes away to human perception. Participants watched a mix of real and generated clips under two tasks, simply understanding the content and actively judging whether each video was artificial. Modern generators make the clips highly realistic, so the researchers expected detection to hinge on subtle physical errors.
What steered looking was not whether a video truly was generated, but whether the viewer believed it was, a perception effect that overrode the actual facts of each clip. Once viewers suspected a fake, their gaze shifted from passive watching into an active hunt for anomalies, scanning for the flaw that would confirm the suspicion. The authors argue this human search behavior should inform both better detectors and more honest generation benchmarks that include real viewers.
Catalog record
| CREATOR | Danqing Shi, Lan Jiang, Katherine M. Collins, Shangzhe Wu, Ayush Tewari, Miri Zilka |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 36 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |