Emotional intelligence in large language models is fragmented across perception, cognition, and interaction
A recent study highlights the fragmented nature of emotional intelligence in large language models (LLMs). The research introduces a new framework called FACET, which evaluates emotional intelligence across various dimensions. Findings indicate that emotional skills in LLMs do not scale uniformly and reveal a common performance bottleneck in emotion recognition.
- ▪The study emphasizes the importance of emotional intelligence in LLMs as they are used in emotionally sensitive areas.
- ▪FACET, the new framework, includes 480 items and is based on the Mayer-Salovey-Caruso four-branch ability model.
- ▪The research categorizes LLM performance into cognitive-dominant, interactive-dominant, and context-dependent profiles.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Artificial Intelligence arXiv:2605.24686 (cs) [Submitted on 23 May 2026] Title:Emotional intelligence in large language models is fragmented across perception, cognition, and interaction Authors:Minghao Lv, Lu Chen, Enchang Zhang, Anji Zhou, Xiaoran Xue, Hanyi Zhang, Fenghua Tang, Zhuo Rachel Han, Mengyue Wu View a PDF of the paper titled Emotional intelligence in large language models is fragmented across perception, cognition, and interaction, by Minghao Lv and 8 other authors View PDF HTML (experimental) Abstract:As large language models (LLMs) are increasingly integrated into emotionally sensitive domains, the structural integrity of their emotional intelligence (EI) becomes a critical frontier for safety and alignment.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
