Why 90% of LLMs Fail in Real-World TV Personalization | OTT Question Time Live 2026 скачать в хорошем качестве

Why 90% of LLMs Fail in Real-World TV Personalization | OTT Question Time Live 2026

A deep dive into the hype surrounding Large Language Models for personalized recommendations, straight from Milan. Professor Paolo Cremonesi, a veteran of recommender system benchmarking spanning two decades, presents findings from years of rigorous testing comparing various techniques against industrial criteria. Discover the surprising reality of LLM performance in production environments and the specific stages within the recommendation pipeline—from metadata enrichment to user interaction—where these powerful models can actually add value or where they significantly fall short compared to established solutions. This analysis challenges common assumptions about artificial intelligence's role in content discovery. This talk was presented at OTT Question Time Live 2026, a VOD Professonial production, on January 30, 2026. #LLMs #recommendersystems #recsys #ai #contentdiscovery #semanticsearch #machinelearning #deeplearning #streamingTV #UX