Analyzing the Correlation Between Hallucinations and Knowledge Conflicts in Large Language Models

Lucrezia Laraspata, Giovanna Castellano, Gennaro Vessio

Hallucinations -- factually incorrect or unverifiable outputs -- remain one of the most challenging limitations of Large Language Models (LLMs), especially in knowledge-intensive t…