Blog/Articles/AI Emotional Intelligence

Could AI Understand Emotions Better Than We Do?

May 22, 2025|Research & Innovation
Research & InnovationMay 29, 2025

Tom Ventura

Founder and CEO of Notle

Share this article

In Brief

A groundbreaking study from the University of Geneva and University of Bern reveals that AI systems can outperform humans in emotional intelligence tests, achieving an 82% accuracy rate compared to the human average of 56%. This finding opens new possibilities for AI applications in education, coaching, and mental health support.

Testing AI's Emotional Capabilities

The research team subjected six leading Large Language Models (LLMs) - including ChatGPT-4, Gemini 1.5 Flash, Copilot 365, Claude 3.5 Haiku, and DeepSeek V3 - to emotional intelligence assessments typically designed for humans. These tests involved emotionally charged scenarios designed to assess the ability to understand, regulate, and manage emotions.

"These AIs not only understand emotions, but also grasp what it means to behave with emotional intelligence," explains Marcello Mortillaro, senior scientist at the UNIGE's Swiss Center for Affective Sciences.

Beyond Simple Understanding

What makes these findings particularly fascinating is that the AI systems demonstrated not just an understanding of emotions, but also the ability to suggest emotionally intelligent behaviors in complex scenarios. For example, when presented with workplace conflict situations, the AI could consistently identify the most appropriate and constructive responses.

The study included scenarios like workplace conflicts, personal relationships, and emotional crisis situations. In each case, the AI systems showed remarkable ability to navigate complex emotional dynamics and suggest appropriate responses.

Creating New Assessment Tools

Perhaps even more impressively, ChatGPT-4 was able to generate new emotional intelligence tests that proved to be as reliable and realistic as traditional tests that took years to develop. These automatically generated tests were validated with over 400 participants, demonstrating reliability comparable to established assessment tools.

  • Tests proved as reliable as traditional assessments
  • Generated in a fraction of the time
  • Adaptable to different contexts and cultures
  • Validated through extensive human testing

Implications for the Future

These findings have significant implications for various fields. In education, AI could help develop emotional intelligence training programs. In coaching, AI systems could assist human coaches in providing emotional guidance. For conflict management, AI could offer unbiased suggestions for resolving emotional disputes.

However, researchers emphasize that AI should augment rather than replace human capabilities. The emotional intelligence demonstrated by AI systems is based on pattern recognition and learned responses, which differs from human emotional intelligence that combines experience, intuition, and empathy.

Tom Ventura

Tom is the Founder and CEO of Notle with a vision for transforming mental healthcare through AI. He founded Notle to bridge the gap between technology and effective mental health support.

Related Articles

AI Chatbots Gaining Traction in Mental Health Support

How AI chatbots are making mental health support more accessible and breaking down barriers.

Read more →

AI Psychometrics: The New Frontier

Exploring how AI systems provide unprecedented consistency in measuring mental health conditions.

Read more →

Experience AI-Enhanced Mental Healthcare

See how our platform combines human expertise with AI capabilities for better mental health outcomes

Request a Demo