Recommendation Systems
📋 What it is
Recommenders predict what you’ll like from what similar users liked — powering feeds and playlists.
🗣️ Coach says
Recommendation systems guess what you’ll enjoy next — videos, songs, products. They spot patterns: "people who liked X also liked Y", and match you to others with similar tastes. They’re powerful and useful, but can also trap you in a "bubble" of only-similar content, so use them thoughtfully.
🧠 Memory hook
Recommenders = "people like you liked this". Powerful + useful, but watch the same-stuff "bubble".
😂 Giggle
The image classifier's favorite genre of music?
Heavy label!
😲 Whoa!
Over 70% of what people watch on some big video sites is chosen by a recommendation AI, not searched for by the user.
✅ Quick check: How does a video app decide what to recommend to you?
Say your answer out loud first — then reveal.
It finds patterns — matching you to users with similar tastes and suggesting what they liked ("people like you liked…").
Recommenders use similarity between users/items.
🧪 Try it! (2 minutes)
Notice your app’s suggestions and ask: why THIS? What pattern about me did it spot?