1 paper
Niclas Lietzow, Danielle Bitterman, Carsten Eickhoff +2
Vision-language models must reconcile visual evidence with memorized world knowledge when the two conflict. How they resolve this conflict shapes the reliability of multimodal syst…