Claim: People have some control over what they pay attention to online, but algorithms also decide what gets put in front of them. They learn from what people watch, search for, like, and stop scrolling to look at.

Evidence: In the example studied, two new accounts received very different recommendations after interacting with different kinds of posts. The users did not search for most of the videos. The app used their earlier behavior to decide what to show them next.

Analysis: This shows that what people see online is not completely their choice. Their choices teach the algorithm what might interest them, but the algorithm then puts more of the same things in front of them. Seeing an idea repeatedly could make it seem more common or more important than it really is.