How do I teach filter bubbles and recommendation algorithms?
How do I teach filter bubbles and recommendation algorithms?
Let students be the algorithm. When they rank videos for imaginary users based on past clicks, they see how recommendation systems learn from behavior and how a feed narrows over time. Then turn to their own feeds and to ways of widening them. This page covers common misunderstandings, classroom moves, a five-lesson plan for Grades 6–9 and ready-made units that carry the material.
View the algorithms unitExplain that recommendation algorithms predict what each user is most likely to watch, click or buy, based on signals such as viewing time, likes, shares and the behavior of similar users. Because they learn from what you already engage with, feeds can narrow into a filter bubble. Teach it by having students act as a recommender for fictional users, compare two contrasting feeds, then design their own algorithm with a different goal and list ways to widen what they see.
Ready-made material
Digital Life Units Bundle, Grades 6–9
6 resources in one download: $89.99 instead of $149.94 bought separately.
The first cover is the main resource; the others are related options. Open any cover for the full contents and price.
Digital Life Units Bundle | Cyberbullying, Influencers, AI, Privacy, Phishing & Algorithms | 6 Units | Grades 6–9
AI & Cybersecurity Units Bundle | How AI Works, Prompting, Phishing, Encryption & Data Privacy | 6 Units | Grades 6–9
Digital Citizenship Activities | Cyberbullying, AI, Social Media & Media Literacy Lessons | Grades 7–10
Middle School Ethics Curriculum | Justice, Conscience, Freedom, Digital Life & Climate | 10 Units | Grades 6–9
Why it is hard
What students assume about their feeds
Most students think their feed is simply what is new or popular. They do not notice that two classmates opening the same app see completely different videos, or that each item was chosen to keep them watching. A common theory is that the phone is listening to their conversations, which hides the simpler truth: viewing time, pauses, likes and follows reveal a lot on their own. Students also assume they choose everything they see, when much of their viewing comes from suggestions they never asked for. Another misunderstanding is that filter bubbles only concern politics. For a thirteen-year-old, a bubble is more likely to be about body image, gaming, a single sport or one style of humor. Students also mix up a filter bubble, where a system narrows what you are shown, with an echo chamber, where people choose to surround themselves with like-minded voices. In class, the topic can become abstract fast, with talk of algorithms nobody can see. It can also turn into a lecture against social media, which students reject. A working model they build themselves, plus a calm look at their own habits, gets much further. How strong bubbles really are is still debated, so present the mechanism rather than a verdict.
What works
Activities that open up the black box
Each activity makes an invisible system visible without needing student accounts. Keep personal feeds private and let students report patterns, not screenshots.
- Act as a recommender: rank ten video titles for a fictional user based on their past clicks.
- Compare two invented feeds for the same app and list what each user is never shown.
- Have students score posts on watch time and likes, then see which rise to the top.
- Ask students to note, privately, the first ten suggestions they see after school and report patterns.
- Design an algorithm with a different goal, such as learning something new, and compare the results.
- Collect concrete ways to widen a feed: searching, following varied accounts, clearing history, taking breaks.
Step by step
Five lessons on feeds and bubbles
The lessons alternate between modeling the system and reflecting on real use. Plan about one class period each.
- Lesson 1: Same app, different feedsShow two invented feeds for one app and ask why they differ. Students list what the app might know about each user and introduce the word recommendation system.
- Lesson 2: Be the algorithmGroups act as recommenders for fictional users using click histories and similar users. They rank suggestions, then notice how quickly each user's options narrow after a few rounds.
- Lesson 3: Goals and signalsDiscuss what platforms aim for, such as time spent, and which signals measure it. Students explain why a system built for attention may not show what is balanced or true.
- Lesson 4: Bubbles and echo chambersSeparate the two ideas with examples from gaming, sport, appearance and news. Students consider which bubbles they might be in and what they could be missing.
- Lesson 5: Redesign the feedGroups design a recommendation rule for a different goal, such as learning, and present it. Each student writes two personal habits for widening their own feed.
Ready-made option
A finished unit on recommendation algorithms
Recommendation Algorithms and Filter Bubbles is part of the Digital Life Units Bundle ($89.99, parts bought separately $149.94), along with units on cyberbullying, influencers, AI and responsibility, data protection and phishing. The unit includes worksheets on four levels, a detailed reading text, a hands-on deep-dive task, slides and full answer keys with a marking guide. On its own it costs $24.99. If your class is more interested in how AI systems work, the AI & Cybersecurity Units Bundle ($89.99) contains the same unit next to How Artificial Intelligence Works and Working with AI Tools and Prompting.
Honest limits
What you will need to adjust
The unit explains how recommendation systems work in general; it cannot show the exact, changing rules of any real platform, and it includes no apps or accounts. Students under platform age limits should not be asked to use social media for the activities, so keep the work on paper or with invented feeds. Update examples to the apps your students talk about. Compare the unit with your state's computer science or media literacy standards before planning. Licences cover your own teaching; check with Teach Lessons by message before team use. A product that does not fit can be exchanged within 7 days for store credit of equal value.
Good to know
More teacher questions
Is my phone really listening to my conversations?
Students raise this every time. Explain that apps can predict interests remarkably well from viewing time, pauses, likes, searches, location and the behavior of similar users, without any listening. Then ask them to test the theory with their own feed patterns. The discussion teaches more about data than a direct answer would.
What is the difference between a filter bubble and an echo chamber?
A filter bubble is created by the system: the algorithm keeps showing you more of what you already engage with. An echo chamber is created by people: users choose to follow and talk mainly with others who agree with them. In practice the two often reinforce each other.
Can I teach this if my students are not on social media?
Yes. Video sites, streaming services, music apps, online shops and game stores all recommend content. Invented feeds and the be-the-algorithm activity work without any accounts, and students who do not use social media often notice patterns more clearly than heavy users, which makes them useful voices in the discussion.
Does the unit align with state computer science standards?
Teach Lessons does not say the unit is aligned with CSTA, ISTE or any state standards. Algorithms, data and their effects on society appear in many middle school frameworks, phrased in different ways. Compare the unit's tasks with your state's documents and decide where it fits.
How do I avoid a lecture against social media?
Stay curious rather than critical. Present recommendation systems as clever tools with goals that are not always the user's goals, and let students judge the trade-offs. Ending with practical habits for widening a feed respects students' choices and gives them something to try right away.
Does this topic belong in computer science or social studies?
Both work. In computer science the focus is on data, signals and how the system makes predictions. In social studies, ELA or advisory the focus shifts to news, opinions and media literacy. Many teachers start with the model in one class and then move to the effects in another.
Turn students into the algorithm
Run the be-the-algorithm game with invented users and see how fast the options shrink. The unit can supply readings, leveled worksheets and answer keys for the rest of the week.
View the algorithms unit