SaTC: Core: Small: Cognitively-Inspired Interfaces and Video Modification Detection

Videos of people created using deep learning tools are becoming more common and more convincing. Such videos can be misleading and even dangerous. New ways are needed to make it clear to a user when a video has been altered or entirely constructed. This project explores different ways of labeling videos to help users better understand the accuracy of videos. This project team is assessing the effectiveness of three different methods for signaling the accuracy of online videos. Two signaling methods that use text or icons to flag videos are inspired by current designs in reference-checking interfaces. A third visual indicator has been developed by the project team. These methods indicate whether a video has been altered. The project team is performing experiments to evaluate the usefulness of visual indicators for this task. The project team also is comparing the different signaling methods to understand which is more helpful for alerting human users that a video has been changed. In one such study, the project team is comparing how each visual indicator changes subjective impressions of videos by measuring whether they change user confidence that a video has been altered. In another study, the project team is comparing the effect of the three signaling methods on memory by measuring how well users remember which videos were modified after being alerted with visual indicators. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.