LLM Hallucinations: Characterization, Quantification, Detection, Avoidance, Mitigation

Explore the critical issue of hallucinations in Large Language Models (LLMs) in this comprehensive 73-minute talk by Vipula Rawte from the AI Institute at UofSC. Delve into the systematic characterization, quantification methods, detection techniques, avoidance strategies, and mitigation approaches for LLM hallucinations. Learn essential frameworks for understanding when and why language models generate false or misleading information, and discover practical solutions to address this significant challenge in AI development and deployment.