Skip to the lesson

This lesson is being checked by experts. How we check lessons

Step 1 of 6Explainer5 min

The guess that felt like mind reading

You open Instagram and the first Reel is exactly the kind of video you’d save. Your Explore page is somehow full of stuff you’d screenshot and send to a friend. It can feel like the app is reading your mind.

It isn’t. It’s making a , a best guess about what you’ll like. Predictions come from patterns.

Most of the AI you use every day was built with . Nobody typed in a rule that says “this person likes skate fail videos.” Instagram studied huge numbers of examples. What you watched all the way through. What you liked. What you scrolled past in half a second. Then it found patterns. Source needed

Think about it

Think of a time Instagram’s feed guessed right, or got it strangely wrong. What had you watched or liked right before that?

In this module, you’ll open the box. You’ll see what goes in and what comes out. You’ll also see why an app can guess so well and still not know what a video is.

Sources3 links behind the facts on this page
  1. Machine learning, explained (Sara Brown, MIT Sloan, April 21, 2021)
  2. Instagram Ranking Explained (Adam Mosseri, Instagram, May 31, 2023)
  3. Learning and Optimization of Implicit Negative Feedback for Industrial Short-video Recommender System (Pan et al., CIKM 2023)