It is well known in the tech world that the ability of humans to recognize social signals is a crucial part of mastering social intelligence. The real question researchers are asking is: can robots learn how to read human social cues and adapt their behavior as a result?
One of the study participants interacting with the robot during the experiment. (Center for Human-Computer Interaction)
Researchers have studied how people react to robots that show faulty behavior compared to perfectly performing robots. This study showed that participants liked the faulty robots way more than the robots that interacted flawlessly.
"Our results show that decoding a human's social signals can help the robot understand that there is an error and subsequently react accordingly," said corresponding author Nicole Mirnig, Ph.D. candidate at the Center for Human-Computer Interaction, University of Salzburg, Austria.
Social robotics is a rapidly advancing field but social robots are not advanced enough to operate without any errors. Most studies in this field are usually based on the idea that robots can perform without flaws.
"Alternatives resulting from unforeseeable conditions that develop during an experiment are often not further regarded or simply excluded," said Mirnig. "It lies within the nature of thorough scientific research to pursue a strict code of conduct. However, we suppose that faulty instances of human-robot interaction are full with the knowledge that can help us further improve the interactional quality in new dimensions. We think that because most research focuses on perfect interaction, many potentially crucial aspects are overlooked."
This research aims to examine a human’s social signal after an interaction with a robot error. In order to do this, the team purposefully programmed faulty behavior into a human-like NAO robot and then participants interact with it. They measured the robot’s likeability, anthropomorphism and perceived intelligence. They analyzed the participant’s reactions when the robots make a mistake. Through video coding, the researchers replicated their findings from earlier studies and found that humans respond to the faulty robot behavior with social signs. The research team used interviews and user ratings and they found that erroneous robots were not seen as less intelligent or anthropomorphic compared to perfect robots. The humans recognized the faulty robot’s mistakes and rated it as more likable than the perfectly performing counterpart.
"Our results showed that the participants liked the faulty robot significantly more than the flawless one. This finding confirms the Pratfall Effect, which states that people's attractiveness increases when they make a mistake," says Mirnig. "Specifically exploring erroneous instances of interaction could be useful to further refine the quality of human-robotic interaction. For example, a robot that understands that there is a problem in the interaction by correctly interpreting the user's social signals could let the user know that it understands the problem and actively apply error recovery strategies."
This discovery has great potential for the social robotics field because they emphasize the importance for robot creators to keep potential imperfections in mind when they are designing robots. Instead of making robots that behave perfectly, designers need to embrace flaws of social robot technology to pave the way for the development of robots that make mistakes and learn from them. This would make the robots more likable and relatable to humans.
The results of this study were published in Frontiers in Robotics and AI.
