Artificial intelligence lags behind humans: it fails to understand social cues.

Despite the significant progress made by artificial intelligence systems in various fields such as engineering, medicine, and space, a new study has shown that these systems still lag behind humans in understanding social cues. According to a study conducted by a research team from Johns Hopkins University, AI systems cannot grasp the "subtext" in social interactions among humans, which could have a significant impact on industries like self-driving cars and robots. In this study, groups of volunteers were asked to watch short video clips lasting no more than three seconds, involving social interactions between people. While volunteers could easily interpret and understand the interactions, AI systems failed to comprehend these social cues as humans did. The clips were analyzed by over 350 specialized AI systems in language, video, and image analysis fields, yet the results showed a significant disparity in interpretation accuracy between humans and machines. The results revealed the failure of artificial intelligence to reach a unified consensus on interpreting social cues present in video clips, while humans arrived at similar answers based on their natural understanding of social interactions. Researcher Leyla Isik, a cognitive science professor at Johns Hopkins University, pointed out that this study sheds light on why artificial intelligence fails to perform some tasks that humans excel at easily, especially those requiring a deep understanding of emotions and unspoken interactions between people. This limited understanding of AI systems may pose challenges in the future, particularly in fields requiring complex interactions between humans and machines. Researchers believe that integrating cognitive science and artificial intelligence could be a solution to overcome this obstacle, helping to develop systems more capable of understanding social cues in interactive environments.