We all lie. “I’m so sorry I was late, there was traffic.” “I’m really sick today and can’t come into work.” “I just saw your text.” How many lies do you tell a day? As technology advances and robots become more ubiquitous, researchers are looking at how humans respond to robot lies.
Researchers from George Mason University conducted a study exploring human perceptions of robot deception. As robots increasingly work alongside humans in fields like healthcare, retail, and household management, ethical questions arise about how robots should behave, particularly regarding honesty. The study, published in Frontiers in Robotics and AI, evaluated how people feel about robots using deception in specific scenarios and how such behavior is justified.
The study involved almost 500 participants and asked participants to evaluate three different types of robot deception across three work environments: medical, cleaning, and retail. The three types of deception studied were external state deception (lying about the outside world), hidden state deception (hiding the robot’s capabilities), and superficial state deception (overstating the robot’s abilities). Each scenario reflected real-world applications of robots and how they might engage in deception during human interactions.
In the first scenario, a robot caretaker lies to a woman with Alzheimer’s, telling her that her deceased husband will be home soon (external state deception). In the second, a cleaning robot films its surroundings without informing a visitor (hidden state deception). In the third scenario, a robot working in a retail store pretends to feel pain while moving furniture to avoid the task (superficial state deception).
After reading these scenarios, the team asked participants to rate the robot’s behavior in terms of approval, level of deception, and justification for the deception. The results revealed that participants most disapproved of hidden state deception—the cleaning robot secretly recording its environment. This type of deception was deemed the most unethical and unjustifiable. Participants were more accepting of the external state deception, where the robot lied to protect the Alzheimer’s patient from emotional pain. The superficial state deception, where the robot pretended to feel pain, was considered manipulative and was also largely disapproved.
Participants frequently blamed developers or robot owners for the hidden and superficial deceptions, indicating that the responsibility for unethical robot behavior lies with those who design or control the robots. However, in the case of external state deception, participants were more lenient, justifying the lie as a means to prevent unnecessary distress.
Members of the team expressed concern over technology’s potential to manipulate users by hiding its true capabilities, noting that companies already use web design and AI chatbots in manipulative ways, urging for regulations to protect users from harmful deception. They emphasized the need for ethical considerations as robots and AI become more integrated into daily life.
The researchers concluded that while this study provides valuable insights, future experiments should involve more realistic human-robot interactions, such as video simulations or roleplays, to better understand how people react to robot deception in real-life settings. This would offer more detailed insights into human perceptions of robotic behavior and inform ethical guidelines for robot-human interactions.
