Search papers, labs, and topics across Lattice.
2
0
4
Systematic misalignments in MLLM-generated captions can be detected with 63.8% accuracy, revealing a critical flaw in image-text pairing that has been largely overlooked.
Current AI models miss critical tumor detections in underrepresented demographics, revealing a hidden bias that could compromise patient outcomes.
