Session 34: Social Media & Misinformation

At a glance: pupils learn how feeds and algorithms choose what they see, why fake news spreads, how to check a viral post with W.A.I.T. (including a reverse image search), what to do about cyberbullying, and how privacy settings, age limits and healthy habits keep social media safe. · Time: 90 minutes (adjust the timings if your period is shorter) · Workbook pages: 6

Learning outcomes

By the end of the session, pupils can:

1. Explain how a feed's algorithm uses signals (watch time, likes, shares, follows) to predict what to show, and what an echo chamber is.
2. Check a viral post with W.A.I.T. (Who posted it? Anyone else reporting it? Image check, Time and date) and decide whether to share it.
3. Respond to cyberbullying with the five steps (don't reply, screenshot, block, report, tell an adult), act as an upstander, and explain privacy settings, the 13+ age limit and healthy habits.

Before the session

• Test the Feed Check lab on the projector PC and a pupil PC; keep the offline pack on USB. No social media account is needed today, and pupils must not sign in to one.
• Prepare viral-flood.jpg on every pupil desktop: an old, widely published news photo of a flood (for example from a well-known news website's report of a past flood in Ghana or elsewhere). Check beforehand that Google's reverse image search finds older copies of it, and note the year and the original story for the answer key.
• Copy the WhatsApp message from Page 5 onto the board or a slide.
• Prepare 6 printed "posts" for the True or Fake? race (3 true, 3 fake), each with a clue: an odd account name, an old date, a story only one unknown page reports, a shouty "share before they delete it!". Keep them non-political.
• Have your school's procedure for reporting bullying and safeguarding worries to hand.

Running order

Key ideas and how to explain them

• The feed and the algorithm: the kenkey seller notices you slow down at the fried fish and puts it at the front. The algorithm watches signals and predicts what will keep you scrolling. It is the "predict" kind of AI you will meet in Session 39.
• Signals: watch time (often the strongest), likes, comments, shares, follows, scrolling past, searches.
• Why it wants you scrolling: adverts pay for free apps. The algorithm's goal is attention, not truth or wellbeing.
• Echo chamber: a room where you only hear your own voice. Seeing a rumour ten times makes it feel true. Pop the bubble: follow a mix, tap "Not interested", look for other sides, use trusted news.
• Misinformation vs disinformation: shared by mistake vs spread on purpose. Both spread through emotion and speed.
• Five kinds of fake: made up, old photo with a new story, edited, deepfake (Session 40), fake account.
• W.A.I.T.: Who posted it? Anyone else reporting it? Image check (reverse image search from Session 12, deepfake signs)? Time and date? If unsure, don't share. This builds on T.H.I.N.K. (Session 33): the T for "True" is what W.A.I.T. helps you check.
• Cyberbullying: repeated hurting, scaring or embarrassing someone online. Don't reply, screenshot, block, report, tell an adult. Never screenshot, save or forward a naked or sexual picture of anyone: report it and tell an adult straight away. Upstanders don't like or share, send a kind message, report and tell.
• Privacy settings: private account, who can message me, location off, approve tags.
• Age limits: most platforms require users to be at least 13. They protect children's data and safety.
• Healthy use: time limits (Digital Wellbeing or Screen Time), screens off an hour before bed, remember the highlight reel, fewer notifications.

The 3D lab

Feed Check is a pretend social media app ("Chatter") on a 3D phone, with the feed shown as a wall of posts behind it. Every account and post is made up. It has three parts, one mission each:

1. Use W.A.I.T. to judge all 6 posts correctly: share the true ones, stop the fakes. Each post has a Who, Anyone else, Image and Time clue (a page created last week, a 2015 flood photo, a deepfake video, "share before they delete it!").
2. Help Esi without replying to the bully (screenshot, block, report, support her, tell an adult), then make all 4 privacy settings safe. Replying or liking the cruel comment makes it worse and means starting again; blocking before the screenshot is stopped. The privacy settings are the workbook's four: private account, messages from friends only, location off, approve tags.
3. Linger on one kind of post until the feed fills up with it, then pop the bubble. After about three "watch to the end"s the wall fills with one topic (an echo chamber); "Not interested" and "Follow a mix" let the other topics back in.

On the projector, demonstrate the first post and talk aloud through your W.A.I.T. check. Pairs then work through the three parts. Ask: "What did the feed think you liked, and how did it know?" The table in Mission 1 on Page 5 records one example of each.

Common mistakes and quick fixes

When things go wrong

• No internet: the Feed Check lab runs offline. For Mission 2, show screenshots of a reverse image search you prepared earlier, and do the W.A.I.T. table on the printed post. Skip step 3 or show a printed help page.
• Not enough PCs: threes for the lab and Mission 2, rotating the mouse each step; the others fill in the W.A.I.T. table.
• A PC freezes or the power goes: the True or Fake? race, the five-step role-play and the scenarios all work on paper.
• The reverse image search finds nothing: that is a teaching point too: a search finding nothing does not prove a photo is real. Go back to W, A and T.

Answer key

Your turn (Page 2): probably more cat and other funny animal videos, because watching three cat videos to the end sent a strong signal (watch time) that he likes them, so the algorithm predicts he will watch more.

Your turn (Page 4): open task. For example: no phones at meal times, screens off an hour before bed, a daily time limit, phones charge outside the bedroom at night.

Mission 1 (Page 5): answers depend on the lab. Look for a named fake post with a W.A.I.T. clue (odd account, nobody else reporting it, an old or edited image, an old date), reporting or blocking instead of replying, and the feed filling up with more of the topic they watched.

Mission 2 (Page 5):

• Who: an unknown page, "Ghana Newz Update 24/7", created last week, with a misspelt name: not a trusted source.
• Anyone else: no trusted news website or the Ministry of Education reports schools closing.
• Image: the reverse image search shows the photo on older pages, from a past flood (give the year you noted). It is not "RIGHT NOW".
• Time: the picture is old; "share before they delete it!" is pressure, not proof.
• Decision: don't share. Kindly tell the group it is false and share the proof.
• How to report: the steps from the platform's own help page, usually: press and hold the post (or tap the ⋯ or Share button) → Report → choose a reason.

Scenario Decisions (Page 6)

1. Forwarding the "lime cures malaria" post straight away: Unwise (a health claim that fails W.A.I.T. and could hurt someone).
2. Fatima doesn't reply, screenshots, blocks and tells her teacher: Wise.
3. Naa laughing along: Unwise (an upstander doesn't join in, sends a kind message, reports and tells).
4. Kojo's reverse image search before sharing: Wise.
5. Abena (11) faking her birth year: Unwise (the 13+ limit protects her; lying removes that protection).

Written Review Questions

1. The app collects signals about what you do (how long you watch, what you like, share, follow, skip and search). From these it predicts what you will want to watch next and puts more of it in your feed, to keep you scrolling.
2. W: Who posted it? A: Anyone else reporting it? I: Image check. T: Time and date. Fake news spreads fast because it uses strong emotion (shock, fear, anger) so people share without thinking, and speed: one tap sends it to many people, who send it on again.

Support and stretch

• Needs more support: do W.A.I.T. as four separate questions, one at a time, with a partner reading aloud. In the True or Fake? race, give them the "Who" clue to look for first.
• Finished early: design a W.A.I.T. poster for the Hub (they can make it in Canva next time), or write a short, kind message a pupil could send in a group chat to correct a fake post without embarrassing anyone.

Evidence for sign-off

The pupil explains the feed loop using "signals" and "predict", completes the W.A.I.T. table for viral-flood.jpg with a correct decision and a real reverse image search, names the five cyberbullying steps in order, and gets at least 4 of the 5 scenarios right. They have completed the Feed Check missions.

Safety and wellbeing

No pupil should sign in to a social media account in the club. Most pupils are under 13 or only just over: don't shame those who already have accounts, but explain the protections and encourage them to set privacy settings with a parent. Talking about cyberbullying can prompt a disclosure. Listen calmly, don't promise secrecy, and follow your school's safeguarding procedure. If a pupil mentions a naked or sexual picture of anyone, never ask them to show or send you the image; follow your safeguarding procedure. Threats, blackmail or scams can be reported, with an adult, to Ghana's Cyber Security Authority (call or text 292). Choose fake-news examples that are not political, religious or about real people in your community, and avoid frightening images. Some pupils may feel low about comparing themselves with others online; keep the highlight-reel message warm and positive.