Are You More Emotionally Intelligent Than an AI Chatbot? This Cognitive Scientist’s Experiment Is the Smartest Test I’ve Seen
A provocative question is gaining traction in the realms of artificial intelligence (AI) and cognitive science: Can machines like ChatGPT truly match—or even surpass—humans in emotional intelligence? To find out, Marcello Mortillaro, a senior scientist at the Swiss Center for Affective Sciences (CISA) at the University of Geneva, conducted a groundbreaking experiment comparing the emotional intelligence of humans and AI chatbots. The results? Surprisingly, AI held its own.
Mortillaro and his team designed a study to test how well large language models could navigate complex emotional scenarios. Using established emotional intelligence assessments commonly used in corporate and research settings, they presented both humans and AI with emotionally charged social situations. For example, one scenario involved a colleague sharing devastating news—being dismissed from their job—and the participant had to choose the most emotionally intelligent response from five options.
The experiment, published in the journal Communications Psychology, tested six large language models, including ChatGPT. The findings were striking: AI chatbots performed similarly to humans in these tests. In some cases, the AI even selected the most emotionally intelligent response as frequently as highly skilled individuals did. This raises a compelling question: Are machines now capable of understanding and responding to emotions as effectively as humans?
While the results challenge the notion that emotional intelligence is a uniquely human trait, they also highlight important limitations. AI’s ability to mimic emotional understanding comes from analyzing vast datasets of human communication—not from genuine feelings or experiences. In real-world scenarios, where non-verbal cues and context play a critical role, the gap between human and machine emotional intelligence may still be significant.
So, what does this mean for the future? Mortillaro’s research suggests that AI could become far more effective in social and emotional contexts than previously imagined. But it also invites caution, as we grapple with what it truly means to be emotionally intelligent in a world where machines are increasingly integrated into human affairs.
The Experiment: Testing Emotional Intelligence in AI vs. Humans
Mortillaro and his team conducted a study published in the journal Communications Psychology, directly comparing the emotional intelligence of AI chatbots with that of humans. They utilized established emotional intelligence tests frequently applied in corporate and research settings. These assessments typically present complex social scenarios, then provide several possible reactions, asking test-takers to select the most emotionally intelligent response.
For example, one scenario reads:
“Your colleague with whom you get along very well tells you that he is getting dismissed and that you will be taking over his projects. While he is telling you the news, he starts crying. He is very sad and desperate. You have a meeting coming up in 10 minutes. What do you do?”
Participants—both humans and AI chatbots—were presented with five possible ways to respond, ranging from pragmatic action to empathetic engagement. The test evaluates the ability to identify, understand, and appropriately respond to the emotions and social complexities at play.
Surprising Results: AI Chatbots Hold Their Own
The experiment ran these tests on six large language models, including generative AI chatbots like ChatGPT. The results, according to Mortillaro, were surprising: the performance of AI chatbots on these tests was not significantly different from that of humans. In some scenarios, AI even selected the most emotionally intelligent response as frequently as highly skilled people did.
This raises compelling questions about the boundaries between human and artificial emotional understanding. Are AI systems, trained on massive datasets of human communication, now able to consistently interpret and respond to emotional cues as well as, or even better than, people themselves?
What Does This Mean for the Future?
The findings challenge the assumption that authentic emotional intelligence is a uniquely human trait. They suggest that, at least in structured, scenario-based tests, AI can demonstrate emotional intelligence that rivals many people’s abilities.
Yet, as the article notes, these results are context-dependent. AI’s apparent emotional intelligence derives from its exposure to vast quantities of human text and its ability to recognize patterns—not from genuine human experience or feelings. In real life, nuances, context, and non-verbal communication often make emotional intelligence far more complex than what can be captured in a standardized test.
Mortillaro’s experiment reveals that large language models like ChatGPT are fast approaching, or even matching, human-like performance in certain tests of emotional intelligence. While this doesn’t mean AI is truly “feeling” or “understanding” in a human sense, it does suggest that future applications for AI in social and emotional contexts could be much broader—and more effective—than previously thought. This research invites both excitement and caution as AI becomes ever more integrated into human affairs, challenging our assumptions about what it means to be emotionally intelligent.
[503] no healthy upstream no healthy upstream