YouTube’s Viewer Rating System Raises Concerns as Faceless Creators Take the Hit

A Kapwing study shows that YouTube is heavily flooded with low‑quality AI slop, with 21 percent of recommended videos for new users and over 40 percent of Shorts shown to children falling into this category. As YouTube faces mounting pressure to clean up the platform, it has begun adjusting its algorithm and testing a viewer rating system that asks users to judge how AI‑generated a video feels. Experts warn this approach is flawed because people are increasingly bad at identifying AI content, and the collected ratings may even be used as training data for YouTube’s own AI models, raising further concerns. These changes are unintentionally harming legitimate faceless creators who produce human‑made videos but do not appear on camera, causing many to lose visibility or hire on‑camera hosts to adapt. The core issue is that YouTube’s algorithm is overcorrecting, punishing the entire faceless format instead of accurately filtering out mass‑produced AI slop.

Editor’s Note: When I first learned that YouTube planned to use its viewers as free trainers for its AI models, it was immediately clear this would lead to serious problems. Many native English speakers can easily recognize AI‑generated speech, and that irritation alone is enough for them to label a video as AI slop. As a result, the system ends up punishing creators who simply speak imperfect English or who happen to have a less pleasant voice.

The irony is that for a large portion of the audience, an AI‑generated voice is actually more pleasant and easier to understand than a heavy accent or unclear pronunciation. Instead of cleaning up low‑quality AI content, YouTube risks harming real human creators who are putting in genuine effort.

Read the full story on Digital Trends →