FriendBench: Benchmarking Dyadic Familiarity Inference in Humans and Multimodal Large Language Models
A new benchmark called FriendBench evaluates whether AI models can infer social familiarity between two people from brief video clips. Top models perform as well as humans in this task.
- FriendBench is the first benchmark to evaluate AI's ability to infer social familiarity from brief video clips of human interactions.
- Top AI models perform as well as humans in detecting familiarity across text, audio, and video modalities.
- The benchmark uses 20-second clips from dyadic ice-breaker conversations to test social inference skills.
- The study compares 26 models from seven companies against human panels across 96 balanced dyads.
Researchers have introduced FriendBench, a benchmark designed to evaluate AI models' ability to infer whether two people in a video clip are already familiar or meeting as strangers. The benchmark uses 20-second clips from dyadic ice-breaker conversations, where only the interaction style reveals the answer. The study compares 26 models from seven companies against matched human panels across 96 balanced dyads, testing performance in text, audio, and video modalities.
The results show that the best-performing AI model achieves accuracy statistically indistinguishable from human performance in every modality. However, the methods by which models and humans reach these conclusions differ, suggesting that AI may be developing alternative strategies for social inference. This benchmark highlights the growing sophistication of multimodal AI systems in understanding nuanced human interactions.
FriendBench is the first benchmark to focus specifically on dyadic familiarity inference, filling a gap in evaluating AI's social intelligence capabilities. The dataset and evaluation framework are designed to be reproducible, enabling further research in this emerging area of AI development.
Provides a new benchmark for evaluating AI models' social intelligence and multimodal understanding.
Highlights the potential for AI systems to better understand human social dynamics, useful for customer service and social media applications.
Demonstrates progress in AI's ability to interpret human behavior, a key differentiator for multimodal AI investments.
Shows how AI is advancing in understanding subtle human interactions.
- dyadic
- relating to or involving two people or parties.
- benchmark
- a standard or point of reference against which things may be compared.
Alibaba unveils its most capable AI model to date, not far behind Moonshot’s in size - WTVB
At Colleges, the AI Boom Means Everyone Wants to Dabble in Computer Science - U.S. News & World Report
Education Notebook: Trine University team to tackle artificial intelligence issues through seven-month program - The Journal Gazette
AI reveals a massive algae boom across the world’s oceans - ScienceDaily
EHR-based AI beckons rapid-response team to head off avoidable in-hospital deaths - HealthExec
SecurityDisrupting a Criminal Scam Operation
OpenAI shut down accounts linked to a Cambodia-based criminal network using ChatGPT for romance and investment scams.
INTERPOL report finds AI linked to more than half of cybercrime in Africa - Interpol
A recent INTERPOL report found that AI is linked to more than half of cybercrime cases in Africa.
EU AI Act Article 50: What the 2026 Transparency Rules Mean for AI Teams
The EU AI Act’s Article 50 introduces enforceable transparency rules starting August 2, 2026, requiring AI teams to document and disclose key system details.
Janesville becomes an AI data center battleground - PBS Wisconsin
Janesville is becoming a key location for AI data centers, with major companies competing for space. This development is expected to bring significant investment and job creation to the area.
Potential US ban on Chinese AI models could cost businesses US$12 billion a year - South China Morning Post
A potential US ban on Chinese AI models could cost businesses up to $12 billion per year, according to a report from the South China Morning Post.
Tech: Casar wants to ban AI superintelligence - Punchbowl News
A U.S. representative has introduced a bill to prohibit the development of AI systems smarter than humans, citing existential risks.