© 2026 ContentWise. All rights reserved.
The Illusion of Progress
ContentWise CTO is in London to show why 90% of LLM projects fail in production
The OTT industry has reached a tipping point.
While Large Language Models (LLMs) promise to revolutionize content discovery, a sobering reality is setting in: most complex AI models look brilliant in a lab but fail the “TV Reality Test.”
On January 28th at 2:40 PM, ContentWise CTO Prof. Paolo Cremonesi will take the stage at OTT Question Time Live 2026 to cut through the generative AI hype.
In a session titled “The Illusion of Progress: Why 90% of LLMs fail in real-world TV personalization,” Paolo will draw on years of peer-reviewed research to explain why complexity often kills production ROI.
What we will explore:
- The Phantom Progress Trap: Why “state-of-the-art” models are often outperformed by well-tuned, simpler algorithms.
- The Latency & Cost Barrier: Navigating the technical unfeasibility of pure LLM architectures at scale.
- Hybrid Intelligence: Why the future of personalization belongs to those who pair semantic depth with scalable engineering.
The future isn’t about choosing between LLMs and traditional Recommender Systems; it’s about orchestrating them to drive real business KPIs. — Prof. Paolo Cremonesi
Related resources
Agent Engine
Move beyond basic recommendations. Unify your engagement strategy with our next-gen personalization engine.
Read more! 
Metadata Lab
Turn your catalog into a discovery engine. Automate and enrich your metadata for better AI performance.
Read more!
UX Engine
Watch viewers grow with your content strategy. Explore our flagship personalization and discovery solution.
Read more!