2026
Mitigating Hallucinations in Large Language Models via Causal Reasoning
AAAI 2026technical
Large language models (LLMs) exhibit logically inconsistent hallucinations that appear coherent yet violate reasoning principles, with recent research suggesting an inverse relationship between causal reasoning capabilities and such hallucinations. However, existing reasoning approaches in LLMs, suc