Search papers, labs, and topics across Lattice.
University of British Columbia
1
0
2
Staggering hallucination rates of up to 86.7% reveal that even advanced language models struggle to recognize non-existent concepts, highlighting a critical vulnerability in AI reliability.
