Three ways to tell if an AI avatar may be hallucinating (IMAGE)
Caption
When a conversational AI avatar in an immersive environment says something false, how should the system alert the user? Researchers tested three techniques for alerting users to hallucinated statements from “embodied” AI agents in virtual reality systems. This image shows examples of each technique: one in which the avatar uses physical gestures to share information with users; one in which icons appear near the avatar; and one in which the screen displays subtitles of the avatar's speech, allowing the system to provide textual cues.
Credit
Xiaoran Yang, NC State University
Usage Restrictions
Credit must be given.
License
Original content