AI Masters Emotional Intelligence Outperforming People
Study Reveals AI's Emotional Prowess
A groundbreaking recent study has shown that six generative AI large language models (LLMs) have surpassed human performance in emotional intelligence tests. This research, published last month in the esteemed peer-reviewed journal Communications Psychology, highlights the rapidly advancing capabilities of artificial intelligence in understanding and processing emotional nuances.
The Contenders AI Versus Humans
The LLMs involved in the study included some of the most advanced models available: ChatGPT-4, ChatGPT-o1, Gemini 1.5 flash, Copilot 365, Claude 3.5 Haiku, and DeepSeek V3. These AI systems were benchmarked against human participants using five distinct emotional intelligence tests. These tests are standard tools for assessing emotional intelligence and included the Situational Test of Emotion Management (STEM), Situational Test of Emotion Understanding (STEU), Geneva EMOtion Knowledge Test (GEMOK), and the Geneva Emotional Competence Test in the workplace.
How AI Fared The Impressive Results
The findings were significant: the AI models achieved an average score of 81% across these tests. In comparison, the average score for human participants was notably lower at 56%. This demonstrates a considerable gap where AI outperformed humans in standardized measures of emotional intelligence.
AI as an Innovator Crafting New Assessments
Intriguingly, the researchers also leveraged the capabilities of AI in the study's design. One of the LLMs, ChatGPT-4, was tasked with generating new assessments and tests that were subsequently utilized for the research. "They proved to be as reliable, clear, and realistic as the original tests, which had taken years to develop," commented Katja Schlegel, one of the co-authors of the study. This indicates AI's potential not just as a subject of study but as a tool in advancing research methodologies.
Expert Insights Understanding AI's Emotional Depth
The study was a collaborative effort by a team of researchers from the University of Geneva (UNIGE) and the University of Bern (UniBE) in Switzerland, along with the Czech Academy of Sciences in Prague. "We chose five tests commonly used in both research and corporate settings," Schlegel added. "They involved emotionally charged scenarios designed to assess the ability to understand, regulate, and manage emotions."
Marcello Mortillaro, another co-author from Geneva's Swiss Center for Affective Sciences, provided further insight into the implications. He stated that the study's results "suggest that these AIs not only understand emotions, but also grasp what it means to behave with emotional intelligence."
Future Implications AI in Socio Emotional Support
The researchers concluded that these sophisticated large language models could be effectively utilized in supporting socio-emotional outcomes. They envision LLMs assisting individuals in gaining new perspectives on emotionally charged situations and potentially aiding them when making significant decisions. This opens up new possibilities for the application of AI in areas requiring emotional understanding and support.
