Georgia Tech HRI study tests how robots can repair trust after lying

Presented at the HRI 2023 conference in Stockholm, a Georgia Tech study had a robot assistant give participants false information in a driving simulation, with 341 people online and 20 in person. Those advised by the robot were far more likely to slow down, suggesting a high level of trust in it. No apology fully restored trust, but once participants learned they had been lied to, an explanation of the robot's reasoning worked best. The authors argue that the public needs to understand robot deception is possible and that policymakers should help regulate it.

Source: Tech Xplore