Recently, Professor Ren Fuji's team, hailing from the School of Computer Science and Engineering (School of Cyberspace Security) at the University of Electronic Science and Technology of China, made a notable splash at ACM CHI 2026. This premier international conference in the realm of human-computer interaction saw the publication of their paper, which subsequently earned them an Honorable Mention Award. The paper, a collaborative effort by Assistant Researcher Deng Jiawen, Master's student Zhang Wentao, Postdoctoral Researcher Jiao Ziyun, and Professor Ren Fuji himself, delves into the reliability and safety challenges posed by conversational AI in intricate scenarios.

As conversational AI becomes increasingly prevalent in areas such as emotional companionship and psychological support, a host of interaction failures tend to emerge during multi-turn dialogues that involve intense emotional expressions and ethical considerations. These failures may manifest as emotional misalignment, a lack of adequate ethical guidance, and unclear response boundaries. To tackle these issues, the research team devised a specialized multi-turn dialogue analysis method tailored for emotionally and ethically charged situations. This method sheds light on the failure patterns of conversational AI in complex interactions, providing invaluable insights for system safety evaluation, interaction diagnosis, and the design of trustworthy AI systems.

ACM CHI stands as one of the most prestigious and influential international conferences in the field of human-computer interaction. This year's edition attracted a staggering 6,730 submissions, with only 1,705 papers making the cut for acceptance. Professor Ren Fuji's team distinguished themselves through the innovative and practical nature of their research, ultimately securing the coveted Honorable Mention Award.