The suggestions on shops, streaming and feeds come from recommendation systems. They learn from what many people do, finding that those who liked certain things tend to like others, and predict what a given person is most likely to want next.
Good recommendations help people discover relevant items among overwhelming choice. The same mechanism can narrow what people see, over-emphasise popular items or chase engagement in unhealthy ways, which has drawn scrutiny. Understanding that these feeds are shaped by predictions, not neutral, helps people use them more deliberately, and has pushed providers toward more control and transparency.