Neural networks excel when games have obvious scores. Recommendations must first decide what “winning” means for human behavior.
Neural networks excel when games have obvious scores. Recommendations must first decide what “winning” means for human behavior.
More from this video
See all →Human moderation after complaints can preserve legal protections. Human approval of everything can…
Optimizing YouTube for shares suppresses private topics and loyal niche audiences that rarely…
YouTube’s ideal black box would optimize ad revenue while judging social, financial, reputational,…
Watch-time optimization cannot recognize truth, quality, context, or safety. Clicks can therefore…
YouTube cannot inspect or repair an opaque recommendation model. It can only change the feedback…
A recommendation system can boost watch time immediately while promoting harmful content and…
YouTube initially gave its recommendation algorithm a single goal: increase watch time.
Creators want YouTube to recommend their videos even when audiences do not want to watch them.
Don't lose this one
A free account saves insights like this to your Boards, and Korva resurfaces them so you actually remember.