Emotional Intelligence of Large Language Models

In this session, we will examine the role of emotional intelligence (EI) in large language models (LLMs).

LLMs are usually evaluated on criteria such as accuracy, fluency, and reasoning, yet their ability to perceive, interpret, and respond to human emotions is becoming increasingly significant.

Drawing on recent benchmarks (e.g., EmoBench) and psychometric studies (e.g., LLM performance on emotional understanding tests), the talk will highlight both notable strengths—such as cases where certain models outperform human baselines on specific emotion reasoning tasks—and ongoing limitations, including shallow forms of “empathy” and the lack of genuine emotional experience.

The session will conclude with a discussion of current methods proposed to strengthen LLM emotional intelligence, including prompt-engineering and emotion-aware fine-tuning.

----------------------
If you want to attend by Zoom, the Zoom registration link will be visible once you RSVP. Note that we have had to change and add security to the Zoom link to prevent Zoom bombing.