Question

Why is my feed full of low-quality AI-generated content?

Vault Verified
Curated Intelligence
Definitive Source
Answer

Because generating it costs almost nothing, recommendation systems reward engagement rather than quality, and the two facts together make mass-produced content economically rational.

The economics. Producing a photograph, article or video used to require time, skill and equipment. Generative tools reduced that to near zero. Anyone can now produce thousands of pieces of plausible-looking content per day. If even a small fraction earns advertising revenue, affiliate clicks or follows that can be sold, the operation is profitable — because the denominator is almost nothing.

Why recommendation systems amplify it. Feeds rank by predicted engagement, not by whether content is true, original or made by a person. AI-generated content optimised to provoke a reaction performs well on exactly the measures the system cares about. The system is working as designed, which is the uncomfortable part.

The recognisable patterns:

Engagement-farming images — implausible scenes, sentimental or emotive setups, elderly people with improbable creations, disasters that did not happen. Comments asking for blessings or reactions, which are the actual product.

Content farms producing high-volume articles on any topic that attracts search or feed traffic.

Repackaged and rewritten versions of other people's work.

Fake product listings and reviews.

Voice-cloned or narrated videos over stock footage.

Why it persists despite complaints. Detection is unreliable and getting harder; labelling depends heavily on self-declaration; moderation is built around harm categories rather than low quality, and most of this content breaks no rule; and it generates engagement, which is not an outcome platforms are strongly motivated to suppress.

What it actually costs. Genuine creators are crowded out, since they cannot match the volume. Trust erodes generally — once enough content is synthetic, people begin doubting real material too, which is the more serious effect. And there is a feedback risk, where models trained on the resulting content degrade.

What you can do: use the "not interested" and block controls, which do influence your own recommendations; follow specific people rather than relying on algorithmic feeds; prefer chronological or curated feeds where available; and check provenance before resharing.

Related Questions