FriendBench: Benchmarking Dyadic Familiarity Inference in Humans and Multimodal Large Language Models
On the map: Multimodal LLMs / vision-language models → MLLM training and visual instruction tuning
Abstract (arXiv)
Reading a social situation often depends on behavior, not words alone. We introduce FriendBench, a benchmark for inferring whether two people are already familiar or are meeting as strangers, from a 20-second clip of a dyadic ice-breaker conversation. Every pair answers the same type of prompt, so only the manner of interaction can reveal the answer. Across text, audio, and video, we compare 26 models from seven companies against matched human panels over 96 balanced dyads. The best model and the human crowd are statistically indistinguishable on accuracy in every modality, but reach it differently: humans stay balanced across the two answers, while the strongest models favor ``stranger.'' This is a difference in effective prior, not in discrimination. Richer channels help both unequally, and only humans gain from visible behavior on top of speech. We release the stimuli, human ratings, and model predictions.
Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in MLLM training and visual instruction tuning, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 10 of its 51 references that are in the map. First posted on arXiv 31 Jul 2026.
Builds on
10 of 51 references.
References · 10
- 2021Calibrate Before Use: Improving Few-Shot Performance of Language Modelscited by 1,392
- 2023Clever Hans or Neural Theory of Mind? Stress Testing Social Reasoning in Large Language Modelscited by 163
- 2020Context-Aware Personality Inference in Dyadic Scenarios: Introducing the UDIVA Datasetcited by 31
- 2017Dual-Glance Model for Deciphering Social Relationshipscited by 21
- 2017A Domain Based Approach to Social Relation Recognitioncited by 18
- 2025SIV-Bench: A Video Benchmark for Social Interaction Understanding and Reasoningcited by 18
- 2020DDRel: A New Dataset for Interpersonal Relation Classification in Dyadic Dialoguescited by 13
- 2025HumanSense: From Multimodal Perception to Empathetic Context-Aware Responses through Reasoning MLLMscited by 9
- 2025Are they lovers or friends? Evaluating LLMs' Social Reasoning in English and Korean Dialoguescited by 3
- 2025AsyReC: A Multimodal Graph-based Framework for Spatio-Temporal Asymmetric Dyadic Relationship Classificationcited by 2
Built on by
Not cited yet.
Similar, not cited
Close in wording, no citation either way
- 2026MeetingToM: Evaluating Multimodal LLMs on Theory-of-Mind Reasoning in Multi-Party Meetings
- 2026GesBench: When and Where Do Multimodal Language Models Perceive Social Meaning?
- 2026GroupMemBench: Benchmarking LLM Agent Memory in Multi-Party Conversations
- 2025SI-Bench: Benchmarking Social Intelligence of Large Language Models in Human-to-Human Conversations
- 2026PIVOTSBench: Evaluating Fine-Grained Interpersonal Relationship Reasoning in Multimodal Large Language Models
- 2026RealCompanion: Benchmarking Human Understanding from Reasoning over Longitudinal Real-World Conversations
- 2026MirrorBench: A Benchmark to Evaluate Conversational User-Proxy Agents for Human-Likeness
- 2021Beyond Goldfish Memory: Long-Term Open-Domain Conversation