How Algorithms Are Quietly Changing What We Think We Like

You watch one video. Then another. Then your feed suddenly seems to understand you.

It is easy to describe this as a clever machine reading your mind. But the more interesting question is not how an algorithm recommends videos. It is what happens when something keeps recommending things to us until our behaviour starts to confirm its own prediction.

The recommendation loop

Most platforms learn from small signals. What you click, how long you watch, what you skip in the first few seconds, what you search for, what you like, share or save. None of these signals says "I enjoy this" in plain words. The system reads them as clues and builds a picture of what is likely to keep you watching.

That picture is then used to choose what to show you next. The goal is usually simple: more attention, more time on the platform, more reasons to come back. Whether you are happier for it is a separate question.

Discovery or creation?

When a feed shows you something you love, it feels like discovery. The algorithm found a thing you already liked but did not know existed.

Sometimes that is exactly what happens. But there is another possibility. If you see something often enough, it becomes familiar, and familiar things feel good. So did the algorithm discover your taste, or did repeated exposure help build it? From the inside, the two feel almost identical.

The feedback loop

The pattern is easy to write down. You watch, the algorithm learns, the algorithm recommends similar things, you watch more, and the algorithm becomes more confident.

Every step makes the next one more likely. The system is not just observing your behaviour. It is shaping the choices that your future behaviour comes from. Research on recommender systems suggests that personalisation can lead to narrower patterns of consumption over time. How strong this effect is, and how much of it deserves the name "filter bubble", is still debated. It is a real question, not a settled one.

When the algorithm gets it wrong

One accidental click can change a feed. You open a video by mistake, or watch something out of curiosity, or look something up for a friend. Suddenly your recommendations tilt in a direction you never chose.

These moments are useful because they make the system visible. For a few days the feed feels slightly off, and you notice that it was making assumptions about you all along.

Are our tastes becoming predictable?

This does not only apply to video. Music, films, news, shopping and even political information now reach us through systems that learn from what we do. The more of our choices pass through them, the more our preferences leave a trail that can be predicted.

Prediction is not the same as control. People still surprise themselves, change their minds and wander off the path. But a system that is right most of the time can start to feel like it knows us, and that feeling can make us less likely to look for anything outside it.

The uncomfortable question

If a system has been influencing what we see every day for years, where does our own preference end and algorithmic influence begin?

There may be no clean line. Taste has always been shaped by what is around us: friends, family, radio, television, the shelf at the local shop. Algorithms are different in scale and in speed, and in the fact that they adjust themselves to each person. That is a good reason to pay attention to them, without panic.

Conclusion

Back to the opening question. Perhaps the most powerful recommendation is not the one that tells us what to watch. It is the one that slowly teaches us what we want to watch.

A small habit helps: now and then, choose something your feed would never suggest. Search for it yourself. It is a simple way to find out which of your tastes are really yours.