Gracenote data shows AI hallucinates 1 in 5 streaming titles completely: Gracenote tested 2,600 titles in 13 countries, finding ungrounded LLMs hallucinated all
A recent test by Gracenote revealed that AI models hallucinate streaming titles at a significant rate, with nearly one in five titles being completely fabricated. The study, which examined 2,600 titles across 13 countries, found that U.S. title accuracy was particularly low, at only 53%. This indicates a widespread issue with the reliability of AI in generating or verifying metadata for media content. AI
IMPACT Highlights significant unreliability in AI-generated metadata for media, potentially impacting content discovery and management.