Recommendations: choosing what appears next
How services rank programmes, products or posts, and how those choices can shape what people see.
Recommendation system case study
What this example shows
Identify the information a recommendation system uses and how ranking affects a person’s choices.
What goes in
Signals such as previous viewing, searches, ratings, purchases, location or what similar users selected.
What the AI system does
A model predicts which available items are most likely to match a goal chosen by the service, such as watch time, relevance or sales.
What comes out
A ranked list of programmes, products, posts or other items shown more prominently than the rest.
What can go wrong
Recommendations can narrow what someone encounters, amplify harmful material or optimise for the service’s interests rather than the person’s wellbeing.
Who is responsible
Services choose the ranking goal and should provide meaningful controls and transparency. People can review history, reset preferences or seek information outside the recommendations.