Algorithmic identity: the profile platforms build without asking

Algorithmic identity is the set of labels a platform assigns to you based on what you do, not what you say. For example, YouTube might tag you as a 'gardening enthusiast' because you watched three pruning videos, even if your profile says 'I hate gardening'. The contrast is crucial: your declared identity is what you type into a bio or settings, while algorithmic identity is inferred from clicks, dwell time, shares, and even mouse movements. In an exam, if a question describes a user being shown ads for vegan products despite never selecting 'vegan' in preferences, that is algorithmic identity at work.

To spot this in a test, look for keywords like 'inferred', 'predicted', 'behavioural profile', or 'calculated label'. A common trick is that platforms use algorithmic identity to personalise recommendations, set risk scores for loans or insurance, or flag content for moderation. One elimination move: if the question mentions a user manually updating their 'interests' list, that is declared identity, not algorithmic. Another hint: algorithmic identity can change without the user knowing, because it updates with every new click or scroll.

A compact way to remember: 'You are what you do, not what you say.' Test yourself by imagining a friend who claims to love cooking but only watches baking shows – the platform's algorithmic identity would label them a baker, not a cook. If an exam item gives you a mismatch between a user's stated preferences and the ads they see, the answer is almost certainly algorithmic identity.