SocialEval: Evaluating Social Intelligence of Large Language Models

TL;DR

SocialEval introduces a comprehensive bilingual benchmark to evaluate the social intelligence of large language models, focusing on both outcome and process aspects of social interactions, revealing current limitations and insights into LLMs' social abilities.

Contribution

The paper presents SocialEval, a novel script-based bilingual benchmark that assesses LLMs' social intelligence through structured narrative scripts and a dual evaluation paradigm.

Findings

LLMs lag behind humans in social intelligence evaluations.

LLMs tend to exhibit prosocial and positive social behaviors.

Representation analysis shows LLMs develop human-like functional partitions.

Abstract

LLMs exhibit promising Social Intelligence (SI) in modeling human behavior, raising the need to evaluate LLMs' SI and their discrepancy with humans. SI equips humans with interpersonal abilities to behave wisely in navigating social interactions to achieve social goals. This presents an operational evaluation paradigm: outcome-oriented goal achievement evaluation and process-oriented interpersonal ability evaluation, which existing work fails to address. To this end, we propose SocialEval, a script-based bilingual SI benchmark, integrating outcome- and process-oriented evaluation by manually crafting narrative scripts. Each script is structured as a world tree that contains plot lines driven by interpersonal ability, providing a comprehensive view of how LLMs navigate social interactions. Experiments show that LLMs fall behind humans on both SI evaluations, exhibit prosociality, and prefer more positive social behaviors, even if they lead to goal failure. Analysis of LLMs' formed representation space and neuronal activations reveals that LLMs have developed ability-specific functional partitions akin to the human brain.