Large Language Models are Highly Aligned with Human Ratings of Emotional Stimuli
This addresses the problem of understanding LLM-human alignment in emotional evaluation for applications like AI proxies, though it is incremental as it builds on existing emotion datasets.
The study investigated how large language models (LLMs) evaluate emotional stimuli compared to humans, finding that GPT-4o aligned closely with human ratings for many emotions (e.g., r ≥ 0.9) but showed weaker alignment for arousal and more homogeneity in responses.
Emotions exert an immense influence over human behavior and cognition in both commonplace and high-stress tasks. Discussions of whether or how to integrate large language models (LLMs) into everyday life (e.g., acting as proxies for, or interacting with, human agents), should be informed by an understanding of how these tools evaluate emotionally loaded stimuli or situations. A model's alignment with human behavior in these cases can inform the effectiveness of LLMs for certain roles or interactions. To help build this understanding, we elicited ratings from multiple popular LLMs for datasets of words and images that were previously rated for their emotional content by humans. We found that when performing the same rating tasks, GPT-4o responded very similarly to human participants across modalities, stimuli and most rating scales (r = 0.9 or higher in many cases). However, arousal ratings were less well aligned between human and LLM raters, while happiness ratings were most highly aligned. Overall LLMs aligned better within a five-category (happiness, anger, sadness, fear, disgust) emotion framework than within a two-dimensional (arousal and valence) organization. Finally, LLM ratings were substantially more homogenous than human ratings. Together these results begin to describe how LLM agents interpret emotional stimuli and highlight similarities and differences among biological and artificial intelligence in key behavioral domains.