AEQ-Bench: Measuring Empathy of Omni-Modal Large Models

TL;DR

AEQ-Bench is a new benchmark designed to evaluate the empathy of omni-modal large models in generating and judging empathetic responses using audio and text cues, revealing strengths and limitations in current models.

Contribution

The paper introduces AEQ-Bench, the first benchmark specifically targeting empathy assessment in omni-modal large models with novel multi-modal evaluation settings.

Findings

Audio-capable models outperform text-only models in empathy tasks.

Models align with human judgments on coarse empathy but struggle with fine-grained paralinguistic cues.

Abstract

While the automatic evaluation of omni-modal large models (OLMs) is essential, assessing empathy remains a significant challenge due to its inherent affectivity. To investigate this challenge, we introduce AEQ-Bench (Audio Empathy Quotient Benchmark), a novel benchmark to systematically assess two core empathetic capabilities of OLMs: (i) generating empathetic responses by comprehending affective cues from multi-modal inputs (audio + text), and (ii) judging the empathy of audio responses without relying on text transcription. Compared to existing benchmarks, AEQ-Bench incorporates two novel settings that vary in context specificity and speech tone. Comprehensive assessment across linguistic and paralinguistic metrics reveals that (1) OLMs trained with audio output capabilities generally outperformed models with text-only outputs, and (2) while OLMs align with human judgments for coarse-grained quality assessment, they remain unreliable for evaluating fine-grained paralinguistic expressiveness.