TY  - RPRT
TI  - Calibrated Ambiguity in Multimodal Language Models: Humans reach for cultural references, while models describe the picture
AU  - Cody Kommers
AU  - Mingrui Ye
AU  - Evelyn Gius
AU  - Daniela Mihai
AU  - Hoyt Long
AU  - Zheng Yuan
AU  - Drew Hemment
PY  - 2026
UR  - https://arxiv.org/abs/2609.12575
ID  - 2609.12575
ER  -