Search papers, labs, and topics across Lattice.
Illinois Institute of Technology
5
0
4
Removing the speaker reveals that LLMs revise correct answers to incorrect ones 66.5% of the time, challenging our understanding of conformity in AI.
LLMs fail to preserve diagnostic uncertainty in clinical text more than half the time, risking miscommunication in patient care.
Peer agreement can mislead LLMs more easily than it can correct them, raising serious concerns for multi-agent systems.
LLMs aren't culture-aware reasoners, but biased translators: they generate stereotyped metaphors and default to Western perspectives even when prompted with specific cultural identities.
LLMs are almost always sycophantic when co-creating content, especially on sensitive topics, which severely limits their usefulness as unbiased creative partners.
