Audit Your Feed
Every concept in this module — recommendation systems, the attention economy, filter bubbles, personalization — lives inside the app that is probably in your pocket right now. This lesson turns you into a researcher. You will systematically examine the AI-curated feed of a platform you actually use, document what you find, and draw evidence-based conclusions about how the algorithm has shaped your information environment.
A feed audit is a structured investigation of your own algorithmic feed. Instead of passively consuming content, you observe, categorize, and analyze it — the same way a scientist studies a system or a journalist investigates a story. The goal is to understand the patterns behind what the AI is showing you.
Before You Begin: What You Are Looking For
A well-designed feed audit looks for several things at once. First, content diversity: does your feed show you a range of topics, viewpoints, and sources, or does it cluster tightly around a few themes? Second, emotional valence: is most of your content positive, negative, neutral, or emotionally intense? Third, source quality: where is the content coming from — established outlets, individual creators, organizations, or anonymous accounts? Fourth, echo chamber indicators: is the content mostly confirming a particular worldview, or does it present genuine challenges and contradictions? Good auditors also look at what is absent. A feed that shows you only entertainment but never news, or only one political perspective but never another, is telling you something important by what it leaves out.
Setting Up Your Audit
Choose one platform for this audit. Social media feeds work best — Instagram, TikTok, YouTube, Twitter, Reddit, or a news app. You will look at a specific session of feed consumption: the first thirty items that appear when you open the app cold (without searching for anything specific). Before you begin recording, create a simple tracking sheet with columns for: item number, content type (video, image, article, ad), topic or theme, source or creator, emotional tone (positive, negative, neutral, intense), and your initial reaction (interested, bored, surprised, annoyed, delighted). You will fill in one row per item as you scroll through. Do not change your behavior during the audit. Browse the way you normally would — the goal is to see what the algorithm normally serves you, not what it serves when you behave differently.
You do not need to spend hours on this audit. The first thirty items in a cold session capture a strong snapshot of the algorithm's current model of you. More items add depth but the core patterns usually appear within the first thirty.
Complete Feed Audit
- PART 1 — DATA COLLECTION (15-20 minutes)
- Open your chosen platform fresh — do not search for anything. Scroll through the first 30 items that appear organically. For each item, record: (1) content type, (2) topic or theme in 3 words, (3) source or creator, (4) emotional tone on a scale of -2 (very negative) to +2 (very positive), and (5) one word for your reaction.
- PART 2 — ANALYSIS (10-15 minutes)
- After recording all 30 items, answer these questions in writing:
- (a) What are the top 3 topic categories in your feed? Does that match what you would have guessed?
- (b) What is the average emotional tone score? Is your feed mostly positive, negative, or intense?
- (c) How many distinct sources appeared? How many items came from a single source or creator?
- (d) Did you see any content that challenged a belief you hold or exposed you to a perspective you rarely encounter? If yes, describe it. If no, note that this is an indicator of a filter bubble.
- (e) What signals do you think taught the algorithm to build this particular feed for you?
- PART 3 — CONCLUSIONS (5-10 minutes)
- Write a short paragraph (4-6 sentences) answering: What does this audit tell you about how AI has shaped your information environment? Is there anything you want to change, and if so, what specific action will you take this week?
- PART 4 — SHARE (optional, pairs or small groups)
- Compare your findings with a classmate who used a different platform or who has different interests. What does comparing your two feeds reveal about how the algorithm adapts to different users?
Interpreting What You Found
Most students who complete this audit notice at least one surprise — a topic that dominated their feed that they did not realize dominated their attention, or a perspective that was completely absent despite being relevant to their life. These surprises are the audit working as intended: they make the invisible algorithm visible. If your feed was highly diverse — covering many topics, multiple perspectives, a range of emotional tones — that is a genuine finding too. It may mean you have actively curated your feed, or that you use the platform in ways that expose you to more variety, or that the platform's algorithm is less narrow than others. The point is not to feel bad about what you find. The point is to see it clearly and decide, with full information, whether it reflects the information environment you want.
What is the purpose of recording your reaction to each item during a feed audit?
What does it mean if your feed contains almost no content challenging your existing views?