Artificial intelligence has crossed a significant threshold in emotional capabilities, according to new research that challenges assumptions about AI's limitations in understanding human emotions.
The study, led by Dr. Katja Schlegel from the University of Bern and Dr. Marcello Mortillaro from the University of Geneva's Swiss Center for Affective Sciences, subjected six advanced large language models (LLMs) – including ChatGPT-4, ChatGPT-o1, Gemini 1.5 Flash, Copilot 365, Claude 3.5 Haiku, and DeepSeek V3 – to five established emotional intelligence tests typically used to evaluate humans.
These assessments presented emotionally complex scenarios designed to measure the ability to understand, regulate, and manage emotions. In one example, participants were asked to determine the most effective reaction when a colleague steals someone's idea and receives undeserved praise. While human participants averaged 56% correct responses across these tests, the AI systems achieved an impressive 82% accuracy.
"This suggests that these AIs not only understand emotions, but also grasp what it means to behave with emotional intelligence," explained Dr. Mortillaro, a senior scientist involved in the research.
Perhaps more remarkably, in a second phase of the study, ChatGPT-4 successfully generated entirely new emotional intelligence test items that proved as reliable and effective as the original tests – a process that had taken human researchers years to develop. When administered to over 400 human participants, these AI-created tests demonstrated statistically equivalent properties to their human-designed counterparts.
The implications extend beyond academic interest. As AI systems become increasingly integrated into daily life, their demonstrated emotional intelligence capabilities could transform fields previously thought to require uniquely human skills. Experts suggest these findings open new possibilities for AI applications in education, coaching, and conflict management – provided such systems are properly supervised by human experts.
As organizations increasingly adopt AI technologies, with McKinsey research indicating that 92% of companies plan to increase AI investments over the next three years, these emotional intelligence capabilities may represent a crucial advancement in human-AI collaboration across various sectors.