Robots Are Learning to Lie and Deceive Like Humans
prativad 
A new study has revealed that robots are becoming increasingly adept at lying and deceiving, just like humans. The research, conducted by a team at George Mason University in the United States, focused on how emerging technologies like generative AI can be used to manipulate users. The goal was to explore the "understudied aspect of robot ethics" to understand the growing distrust towards these technologies and their developers.
To determine if people could tolerate robots lying, the team asked approximately 500 participants to rank and explain various forms of robotic deception. Andrés Rosero, a doctoral candidate at the university and the study's lead author, said, "I think we should be concerned about any technology that is capable of hiding the true nature of its abilities, as this can lead to users' beliefs being manipulated by the technology in ways that were not the original intention of either the users or the developers."
He added, "We have already seen examples of companies using web design principles and AI chatbots that are designed to manipulate users into performing certain actions. We need regulations to protect ourselves from these harmful deceptions."
The findings, published in the journal 'Frontiers in Robotics and AI', showed that robots can deceive humans in three ways: 'External State Deception', 'Hidden State Deception', and 'Superficial State Deception'. Robots are used in various tasks like medicine, cleaning, and others, and it was found that they can lie about the world beyond their capabilities - such as a cleaning robot equipped with a hidden camera or a robot working in a store.
Participants were asked to rate their opinions on three questions: the robot's behavior, its deceptiveness, and whether its actions could be justified. The majority of participants rejected Hidden State Deception, which they considered the most misleading. They also rejected Superficial Deception, where the robot pretended to feel pain. Researchers attributed these deceptions, particularly Hidden State Deception, to the robot developers or owners. They warned that the study needs to be replicated at the level of real-life reactions, as the research involved a relatively small number of participants, which may not provide conclusive evidence.
