TECH NEWS – A new study suggests that ChatGPT-4 and other advanced AI models are capable of surpassing humans in emotional intelligence, potentially opening up new opportunities in education.

Artificial intelligence, and in particular large language models (LLMs) with advanced capabilities, have exceeded expectations. Tech giants are increasingly focused on developing more efficient AI models and leveraging this technology to handle many everyday tasks on behalf of consumers. A recent study evaluated how these models perform when it comes to emotional intelligence, and the results were surprising.

Researchers from the University of Geneva and the University of Bern recently conducted a study that explored the empathetic abilities of AI. Published in the journal Communications Psychology, the study produced intriguing results. It found that generative AI models like ChatGPT not only demonstrated emotional intelligence but also outperformed humans in related tests. The study examined six leading LLMs: ChatGPT-4, ChatGPT-o1, Gemini 1.5 Flash, Claude 3.5 Haiku, Copilot 365, and DeepSeek V3.

The study included five emotional tests commonly used in academic and professional settings to measure emotional understanding, regulation, and management. These scenarios were realistic and emotionally charged, intended to see how the models responded to given emotional contexts. The results left researchers baffled: all LLMs significantly outperformed human participants. Going a step further, the researchers even asked ChatGPT-4 to generate new test scenarios, which were then validated by human participants, and the results were consistently impressive, showing that the AI models have a strong contextual understanding.

“LLMs can not only identify the best answer among various options but also create new scenarios that align with the intended context. This confirms the idea that LLMs—such as ChatGPT—possess emotional knowledge and can engage in reasoning about emotions,” said Marcello Mortillaro, one of the senior researchers.

These results are crucial, especially as technology continues to surpass expectations in domains traditionally reserved for humans. This could have wide-ranging implications for complementing human skills in areas like conflict resolution and coaching.

Source: WCCFTech, Nature