People Are More Willing to Lie to AI Than Humans, Study Finds
Consumers are more likely to lie or act unethically toward chatbots and AI agents than toward human customer-service workers. This is because interactions with AI involve less social pressure and fear of judgment, a phenomenon researchers call “anticipatory face loss.” People feel less embarrassment or risk of social disapproval with AI, since AI is perceived as less socially aware and less capable of judging. As a result, unethical behaviors such as lying, claiming false discounts, or exploiting pricing errors are more common when dealing with AI systems. However, dishonest behavior decreases if the AI agent is perceived as more competent or makes simulated eye contact. As companies increasingly use AI for customer support roles, understanding these dynamics is important. Related research shows that people respond more positively to humanoid robots with moderate human features, such as facial expressions and eye movement, but find highly realistic robots unsettling. Adding human characteristics to robots enhances trust, comfort, and willingness to interact.
