Research from Imperial College London's Dyson School of Design Engineering, reveals that humans instinctively defend AI bots against social ostracism, mirroring behaviours typically reserved for human-to-human interactions. The study, published last month in Human Behavior and Emerging Technologies, involved 244 participants aged 18 to 62 playing a virtual ball game called "Cyberball."
The study utilised Cyberball, a simple ball-tossing game, to observe human reactions when AI bots were deliberately excluded by other players. Researchers found that participants consistently attempted to compensate for unfair treatment by preferentially passing the ball to excluded AI bots, with older participants showing heightened sensitivity to perceived unfairness.
While this human tendency to treat AI as social beings could benefit workplace collaboration, they caution against using virtual agents as substitutes for human relationships or as advisors on physical and mental health matters.
The study's authors acknowledge that Cyberball interactions might not fully represent real-world scenarios, where humans typically interact with AI through written or spoken language, via chatbots or voice assistants. To address this limitation, the research team is designing new experiments examining face-to-face conversations with AI agents in various settings, including laboratory and casual environments.