AI Knows What You'll Pick Next! 🔮
How recommendation AI predicts your next favourite song, show, or product
AI Knows What You'll Pick Next! 🔮
You're Totally Predictable (And That's Useful!)
This sounds a bit creepy, but it's true: your tastes and behaviour follow patterns. And AI is EXTREMELY good at spotting those patterns to predict what you'll like.
🎵 How Spotify Discover Weekly Really Works
Every Monday, 400 million Spotify users get a unique 30-song playlist they've never heard — but will probably love.
Here's the actual process:
Step 1: Analyse YOUR listening history
↓ What genres, artists, moods, energy levels do you love?
Step 2: Find your "music twin"
↓ Which of 400M users has MOST similar taste to you?
Step 3: Borrow from your twin
↓ What do THEY love that you haven't heard yet?
Step 4: Check the song itself
↓ Does it match your preferred tempo, key, mood?
Step 5: Build your Discover Weekly! 🎵
📺 Netflix: 80% of What You Watch is AI-Recommended
Think about the last show you watched on Netflix. Did you search for it, or did Netflix suggest it?
Studies show 80% of Netflix watching comes from AI recommendations!
Netflix has 8,000+ categories, including ridiculous ones AI invented:
- "Critically-Acclaimed Emotional Underdog Movies"
- "Mind-Bending Sci-Fi from the 1980s"
- "Feel-Good Movies Starring Strong Female Characters"
AI creates categories that match YOUR taste perfectly — not categories that exist in the real world!
🛒 Two Types of Recommendation AI
Collaborative Filtering (the "People Like You" approach)
You like: Minecraft, Roblox, coding tutorials
Other users with same taste also like: board game videos, 3D printing
→ AI suggests board game content to you!
(Even though you never searched for it)
Content-Based Filtering (the "Similar Stuff" approach)
You loved: Avatar: The Way of Water
Features: Action, sci-fi, stunning visuals, long runtime
→ AI finds more films with those same features!
Most big platforms combine BOTH types!
💡 Did You Know?
TikTok's "For You Page" algorithm analyses 10+ signals per second while you watch: how long you watched, did you replay, did you share, what time of day, what other videos you watched before... It's incredibly accurate and incredibly addictive! 📱
Quick check
What percentage of Netflix viewing comes from AI recommendations?