Search papers, labs, and topics across Lattice.
Seoul National University Hospital
2
0
3
Misleading textual context can cause VLMs to switch predictions incorrectly up to 78% of the time, revealing a critical vulnerability in clinical AI systems.
LLMs may encode dangerous biases and inaccuracies, revealing a critical need for interpretability in medical applications.
