AVMeme Exam: A Multimodal Multilingual Multicultural Benchmark for LLMs' Contextual and Cultural Knowledge and Thinking
Xilin Jiang · Qiaolin Wang · Junkai Wu · Xiaomin He · Zhongweiyang Xu · Yinghao Ma · Minshuo Piao · Kaiyi Yang · Xiuwen Zheng · Riki Shimizu · Yicong Chen · Arsalan Firoozi · Gavin Mischler · Sukru Samet Dindar · Richard Antonello · Linyang He · Tsun-An Hsieh · Xulin Fan · Yulun Wu · Yuesheng Ma · Chaitanya Amballa · Weixiong Chen · Jiarui Hai · Ruisi Li · Vishal Choudhari · Cong Han · Yinghao Aaron Li · Adeen Flinker · Mounya Elhilali · Emmanouil Benetos · Mark A. Hasegawa-Johnson · Romit Roy Choudhury · Nima Mesgarani
AVMeme Exam is a human-curated multilingual and multicultural benchmark of audio-visual Internet memes, showing that multimodal LLMs struggle with textless media and with contextual, cultural, and pragmatic thinking beyond content.
摘要
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.