Comcast paper: LLM user profiles only beat aggregates for exploratory viewers
_reachsumit · x · 2026-10-01
A Comcast team evaluated aggregate vs. LLM-generated user profiles on a production streaming recommender in a 2×2 design crossing representation type with contextual scope.
Key findings:
- Aggregate profiles consistently win for habitual viewers (80% of users); LLM profiles are stronger only for exploratory users whose behavior diverges semantically from history.
- LLM-generated profiles show a popularity-attractor effect: slightly higher within-list diversity but substantially lower catalog coverage and reduced novelty.
The takeaway: LLM-inferred context isn't universally better—its value depends on the user's consumption regime, giving practitioners an evidence base for profile selection in production.
More from Research
- Protein binding folding models face data starvation after PDB has been fully juiced — anshulkundaje · 2026-10-01
- Stanford's UniEvo-VL Self-Distillation Lifts Qwen-image GenEval From 0.747 to 0.808 — stanfordnlp · 2026-10-01
- NVIDIA's Mid-Harness Scales Actions at the Model-Harness Boundary, Lifting TerminalBench Pass@1 to 68.03% — nvidia · 2026-10-01
- Amazon's SMART Self-Evolving Multi-Agent System Tops All 15 Subtitle Arena Directions, Cuts Penalty 6.9% — amazon · 2026-10-01
- Meta's Loop Scaling Laws: Sparsity Gives ~3x Active-Param Efficiency, Recurrence ~2x on Reasoning — facebook · 2026-10-01
- AI Protein Design Still Can't Solve Binder Prediction and the Age-Old Docking Problem, Researcher Explains — anshulkundaje · 2026-10-01