AVMeme Exam: A Multimodal Multilingual Multicultural Benchmark for LLMs' Contextual and Cultural Knowledge and Thinking

TL;DR

The AVMeme Exam benchmark assesses multimodal large language models' ability to understand internet audio-visual memes in cultural and contextual terms, revealing current models' limitations in perceiving deeper meaning beyond surface content.

Contribution

This paper introduces AVMeme Exam, a novel benchmark for evaluating multimodal models' understanding of cultural, contextual, and emotional aspects of internet memes across audio and visual modalities.

Findings

Models perform poorly on textless music and sound effects.

Models struggle with contextual and cultural understanding.

Current models show a significant gap in human-aligned multimodal intelligence.

Abstract

Internet audio-visual clips convey meaning through time-varying sound and motion, which extend beyond what text alone can represent. To examine whether AI models can understand such signals in human cultural contexts, we introduce AVMeme Exam, a human-curated benchmark of over one thousand iconic Internet sounds and videos spanning speech, songs, music, and sound effects. Each meme is paired with a unique Q&A assessing levels of understanding from surface content to context and emotion to usage and world knowledge, along with metadata such as original year, transcript, summary, and sensitivity. We systematically evaluate state-of-the-art multimodal large language models (MLLMs) alongside human participants using this benchmark. Our results reveal a consistent limitation: current models perform poorly on textless music and sound effects, and struggle to think in context and in culture compared to surface content. These findings highlight a key gap in human-aligned multimodal intelligence and call for models that can perceive contextually and culturally beyond the surface of what they hear and see. Project page: avmemeexam.github.io/public