Source: The Verge
YouTube is automating its recommendation engine—letting users generate custom feeds on-demand rather than relying on algorithmic curation. This signals that algorithmic feed fatigue is eroding watch time, and YouTube would rather own the solution than lose viewers to ChatGPT or competitors. The move trades long-term engagement metrics (watch history, dwell time, implicit preference signals) for immediate satisfaction, which only makes sense if YouTube believes satisfied users will generate more ad impressions or premium conversions than the algorithmic treadmill.