A 96-dyad benchmark with matched human ratings shows multimodal LLMs match human crowd accuracy on familiarity inference but rely on a stranger response bias and underuse visible behavior.
PIVOTSBench: Evaluating Fine-Grained Interpersonal Relationship Reasoning in Multimodal Large Language Models
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
abstract
Humans possess an innate ability to understand fine-grained interpersonal relationships, which is central to everyday social interactions. Although such reasoning is inherently multimodal, it remains largely unexplored by existing multimodal large language models (MLLMs). To address this gap, we introduce PIVOTS, the first benchmark built from Social-IQ 2.0 and YouTube data to evaluate MLLMs' ability to predict bidirectional interpersonal relationship dimensions grounded in established psychology research. In addition, PIVOTS includes auxiliary tasks that assess models' ability to identify and leverage the critical visual cues underlying such predictions. We evaluate both proprietary and open-source MLLMs and conduct detailed ablation studies to analyze the effects of visual modalities and explicit social role information in conversational utterances. We further examine how joint and pairwise prediction settings benefit MLLMs in scoring bidirectional PIVOTS dimensions. Project page and resources: https://flynnzhangsx.github.io/PIVOTSBench/ .
fields
cs.CL 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
FriendBench: Benchmarking Dyadic Familiarity Inference in Humans and Multimodal Large Language Models
A 96-dyad benchmark with matched human ratings shows multimodal LLMs match human crowd accuracy on familiarity inference but rely on a stranger response bias and underuse visible behavior.