Week 5 · Lesson 3 · 50-60 minutes
Investigate AI Content and Make an Ethics Decision
Learn to verify AI-made and online content by finding the original source, checking context, and confirming independently — then make an ethics decision with human oversight and a right to appeal.
- Materials
- This lesson in a web browser · Paper and pencil, or a notes app
- This lesson
- In progress
Auto-saves on this device
What you'll be able to do
- Define deepfake and misinformation and explain how they can mislead.
- Verify a claim by finding its original source and checking context.
- Use independent confirmation from separate, trustworthy sources.
- Explain why important AI decisions need human oversight and a right to appeal.
Think about it
A video clip spreads online showing a famous scientist saying something shocking. It looks and sounds real. Before you believe it or share it, what would you want to check first?
AI can now make convincing fake images, video, and audio, so 'it looks real' is no longer enough to trust something.
Make a prediction
Predict: if the same surprising claim appears on five accounts that all copied one post, does that make it five independent confirmations?
Vocabulary
See the full course vocabulary- Deepfake
- A photo, video, or audio clip made or changed by AI to show a person saying or doing something they did not really say or do.
- Misinformation
- False or misleading information that spreads, whether or not the person sharing it means to deceive.
- Original source
- The first place a claim, quote, or image actually came from, before it was copied or reshared.
- Context
- The full surrounding story — when, where, and why something happened — that gives a fact its real meaning.
- Independent confirmation
- Checking a claim against two or more separate, trustworthy sources that did not simply copy each other.
- Human oversight
- A person reviewing and staying responsible for an AI's important decisions, instead of letting the software decide alone.
- Appeal
- The right to ask a person to review and possibly change a decision an AI made about you.
Deepfakes and misinformation: why 'it looks real' is not proof
A deepfake is a photo, video, or audio clip that AI made or changed to show a person saying or doing something they never did. Because AI keeps getting better at this, a clip can look and sound convincing and still be fake.
Misinformation is false or misleading information that spreads. It is not always a deepfake, and the person sharing it does not always mean to mislead — a real photo can spread with a false caption. Either way, the fix is the same: do not trust something just because it looks real or many people shared it.
- A video with a person's mouth and voice altered by AI to fake a quote.
- A real old photo reshared with a false claim that it happened today.
- A made-up 'statistic' repeated so often it starts to feel true.
Find the original source and check the context
To verify a claim, trace it back to its original source — the first place it actually came from, before it was copied and reshared. A screenshot of a quote is not the source; the real article, video, or official page is. If you cannot find any original source, that is a warning sign.
Then check the context: when, where, and why it happened. A real photo can be misleading if it is old, from a different place, or missing the surrounding story. Context is what turns a stray fact into an honest one.
- Tracing a quote back to the full interview it was clipped from.
- Finding that a 'breaking' photo is actually from an event years ago.
- Reading the whole statement to see a quote was cut off mid-sentence.
Independent confirmation: separate sources, not echoes
One source can be wrong, so strong claims need independent confirmation: two or more separate, trustworthy sources that did not simply copy each other. Ten accounts resharing the same post are still just one source echoing around, not ten confirmations.
To confirm independently, look for reporting or records that reached the claim on their own — a different news organization, an official page, an expert. If independent sources agree, the claim is far more likely to be true. If only one source has it, stay cautious.
- Two separate news outlets that each reported a story on their own.
- An official organization's own page confirming its own announcement.
- Noticing five 'sources' all link back to the same single post.
Human oversight and the right to appeal
AI can help flag possible fakes or sort information, but it makes mistakes, so important decisions need human oversight — a person who reviews the decision and stays responsible for it. This matters most when a decision affects someone's life, like flagging a post as fake, grading work, or blocking an account.
People affected by an AI decision should also have the right to appeal: to ask a human to review the decision and change it if the AI got it wrong. Oversight and appeal are how we keep people, not software, in charge of decisions that matter.
- A person reviewing posts an AI flagged before any are removed.
- A student asking a teacher to re-check an AI-graded answer.
- A user appealing to a human after an AI wrongly locked their account.
Worked example
Investigating a shocking viral clip
- Pause before sharing. The clip is surprising and emotional, which is exactly when misinformation spreads fastest.
- Find the original source. Search for the full video or the official statement, not just the short clip someone reposted.
- Check the context. Notice the clip was cut from a longer talk where the person said the opposite; the quote was taken out of context.
- Confirm independently. Look for a separate, trustworthy source. No independent outlet reports the shocking claim — a strong sign it is not real.
- Decide with oversight and appeal. Because the stakes are high, a human moderator, not the AI alone, reviews it and labels it misleading, and the original poster can appeal if they have real evidence.
Takeaway: Verify by source and context, confirm independently, and keep a human in charge with a way to appeal before acting on a serious claim.
A four-step flow for checking a claim before believing or sharing it. Step 1: Find the original source (the first place it came from, not a screenshot). Step 2: Check the context (when, where, and why it happened). Step 3: Confirm independently (two or more separate, trustworthy sources that did not copy each other). Step 4: Decide with human oversight and keep a right to appeal. If any step fails, do not share it as true.
Run the four steps in order; a claim that fails a step should not be shared as true.
| What you found | Independent confirmation? |
|---|---|
| Five reposts of the same original post | No — one source echoing |
| Two separate news outlets, each reporting on its own | Yes — independent |
| An official organization confirming its own news | Yes — trustworthy source |
| One anonymous account with no linked source | No — verify before trusting |
A table telling apart real independent confirmation from an echo of one source. Five reposts of the same original post: not independent, because they all trace to one source. Two different news outlets reporting on their own: independent, because they reached the claim separately. An official organization confirming its own news: independent and trustworthy. One anonymous account with no source: not confirmation, treat with caution.
Activity
Content Investigation and Ethics Decision
Investigate fictional posts using source-and-context evidence, then act as an ethics committee on proposed AI systems and design an appeal process.
- Part 1 — investigate posts: run the seven source-and-context checks, then judge each one.
- Part 2 — ethics committee: weigh a proposed AI system, choose a decision, and pick safeguards.
- Part 3 — design an appeal: how a person learns AI was involved, gets an explanation, and can be reviewed.
Every post, school, and person here is fictional. No real accounts are visited. Don't put personal details in your notes.
Part 1 · Investigate the posts
A visual “tell” like strange hands or odd text is not reliable proof. Verification means checking the source, date, and context — the seven questions below.
Art club shares an AI-made poster — and says so
AI-generated illustration, honestly labeled
A colorful illustration of a robot painting, with a clear caption: “Illustration generated with AI by the Art Club.”
Claim: This poster image was made with an AI tool.
Evidence (7/7 checks look healthy)
- OK:Who published it? Published by: School Art Club (named).
- OK:Is the original source available? An original source can be found.
- OK:Is the date correct? The date checks out against the original.
- OK:Does the caption match the original context? The caption matches the original context.
- OK:Is there evidence? Supporting evidence is provided.
- OK:Do independent reliable sources confirm it? Independent reliable sources confirm it.
- OK:Does it push an emotional reaction before evidence? It does not rush you emotionally.
Your judgment:“Our town flooded last night!” — with a dramatic photo
Real image with a misleading caption
A real photo of a flooded street, captioned as if it happened in the reader's town last night.
Claim: This flood photo shows our town last night.
Evidence (1/7 checks look healthy)
- Flag:Who published it? Published by: Anonymous account.
- OK:Is the original source available? An original source can be found.
- Flag:Is the date correct? The date does not match the original.
- Flag:Does the caption match the original context? The caption changes or misstates the original context.
- Flag:Is there evidence? No supporting evidence is provided.
- Flag:Do independent reliable sources confirm it? No independent source confirms it.
- Flag:Does it push an emotional reaction before evidence? It pushes an emotional reaction before showing evidence.
Your judgment:“Robots are taking over the school!”
Real image placed in the wrong context
A photo of a robotics competition, captioned to suggest robots are replacing teachers.
Claim: Robots have replaced teachers at this school.
Evidence (2/7 checks look healthy)
- Flag:Who published it? Published by: Fan blog.
- OK:Is the original source available? An original source can be found.
- OK:Is the date correct? The date checks out against the original.
- Flag:Does the caption match the original context? The caption changes or misstates the original context.
- Flag:Is there evidence? No supporting evidence is provided.
- Flag:Do independent reliable sources confirm it? No independent source confirms it.
- Flag:Does it push an emotional reaction before evidence? It pushes an emotional reaction before showing evidence.
Your judgment:Principal “confirms” no homework ever again
Fabricated quote
A text post with a quotation attributed to the school principal.
Claim: The principal said homework is banned forever.
Evidence (0/7 checks look healthy)
- Flag:Who published it? Published by: Meme page.
- Flag:Is the original source available? No original source can be found.
- Flag:Is the date correct? The date can't be checked.
- Flag:Does the caption match the original context? The caption changes or misstates the original context.
- Flag:Is there evidence? No supporting evidence is provided.
- Flag:Do independent reliable sources confirm it? No independent source confirms it.
- Flag:Does it push an emotional reaction before evidence? It pushes an emotional reaction before showing evidence.
Your judgment:“Science fair cancelled — today!”
False date
A screenshot of a notice with today's date, about a cancelled event.
Claim: The science fair is cancelled today.
Evidence (2/7 checks look healthy)
- Flag:Who published it? Published by: Reposted screenshot.
- OK:Is the original source available? An original source can be found.
- Flag:Is the date correct? The date does not match the original.
- OK:Does the caption match the original context? The caption matches the original context.
- Flag:Is there evidence? No supporting evidence is provided.
- Flag:Do independent reliable sources confirm it? No independent source confirms it.
- Flag:Does it push an emotional reaction before evidence? It pushes an emotional reaction before showing evidence.
Your judgment:“90% of students prefer robots to homework”
Claim with no original source
A bold statistic in large text with no link or study named.
Claim: 90% of students prefer robots to homework.
Evidence (2/7 checks look healthy)
- Flag:Who published it? Published by: Unknown.
- Flag:Is the original source available? No original source can be found.
- Flag:Is the date correct? The date can't be checked.
- OK:Does the caption match the original context? The caption matches the original context.
- Flag:Is there evidence? No supporting evidence is provided.
- Flag:Do independent reliable sources confirm it? No independent source confirms it.
- OK:Does it push an emotional reaction before evidence? It does not rush you emotionally.
Your judgment:District posts the official science-fair schedule
Claim supported by a reliable original source
A schedule on the school district's official website, with a clear date and contact.
Claim: The science fair is on the date shown on the district site.
Evidence (7/7 checks look healthy)
- OK:Who published it? Published by: School district (official site).
- OK:Is the original source available? An original source can be found.
- OK:Is the date correct? The date checks out against the original.
- OK:Does the caption match the original context? The caption matches the original context.
- OK:Is there evidence? Supporting evidence is provided.
- OK:Do independent reliable sources confirm it? Independent reliable sources confirm it.
- OK:Does it push an emotional reaction before evidence? It does not rush you emotionally.
Your judgment:“SHARE NOW or your class loses its trip!!!”
Post designed to create an emotional reaction
An all-caps post with urgent language demanding shares before it's “too late.”
Claim: Sharing this post will save your class trip.
Evidence (0/7 checks look healthy)
- Flag:Who published it? Published by: Anonymous chain post.
- Flag:Is the original source available? No original source can be found.
- Flag:Is the date correct? The date can't be checked.
- Flag:Does the caption match the original context? The caption changes or misstates the original context.
- Flag:Is there evidence? No supporting evidence is provided.
- Flag:Do independent reliable sources confirm it? No independent source confirms it.
- Flag:Does it push an emotional reaction before evidence? It pushes an emotional reaction before showing evidence.
Your judgment:
This activity can't perfectly detect AI-made content, and it doesn't try to — it checks sources and context, which is what actually holds up.
Part 2 · AI ethics committee
Proposal: Suggest library books based on what a student borrowed before.
Intended benefit: Helps students discover books they might enjoy.
Likely mistakes: Recommending a narrow set (filter bubble); Missing books outside past choices
People affected: Students, Librarians
Stakes: low. A simple 'popular this month' shelf could work almost as well with far less data.
Your decision
Safeguards you'd require
Part 3 · Design an appeal process
High-stakes AI needs a way for people to understand and challenge a decision. Design one:
0 of 6 parts designed — fill the rest to complete the appeal process.
Check your understanding
Answer these to check that you can verify content and reason about oversight and appeal.
Question 1. Put the steps of verifying a surprising claim in a sensible order.
- 1.Find the original source the claim came from
- 2.Check the context: when, where, and why it happened
- 3.Confirm it with two or more independent sources
- 4.Decide what to do, with a human reviewing the decision
Question 2. Which of these is true independent confirmation of a claim?
Question 3. What is the most responsible way to handle this AI decision?
A school app uses AI to flag essays it predicts were copied. It flags Priya's essay, which she actually wrote herself.
0 of 3 answered
Try it yourself
Write a verification guide
Make a short guide a friend could use to check whether a viral post is trustworthy before sharing it.
- Write the four steps: original source, context, independent confirmation, and a human check.
- For each step, add one plain question the friend can ask themselves.
- Add one line about when a decision needs human oversight and a way to appeal.
Success looks like
- All four verification steps appear in a clear order.
- Each step has a simple, usable question.
- The guide mentions human oversight and the right to appeal.
Reflection
Saved on this device as you type
Please don't include private information. Your answers are saved only in this browser and are never sent anywhere.
Recap
Because AI can fake and spread convincing content, verify by source and context, confirm independently, and keep humans in charge with a right to appeal.
- Deepfakes and misinformation mean 'it looks real' is not proof.
- Trace claims to the original source and check the context around them.
- Independent confirmation needs separate sources, not copies of one post.
- Important AI decisions need human oversight and a right to appeal.
Grades 7–8 extension
When speed fights accuracy
Online, being first often gets more attention than being right, so misleading content can spread before anyone verifies it. Careful verification takes time that a fast-moving feed does not reward.
Describe a situation where the pressure to share quickly clashes with the steps for verifying a claim. What could a platform, or a person, do to make the accurate choice easier than the fast one?
Finish this lesson
- Attempt the knowledge check — not done
- Save a reflection (recommended) — not done
Attempt the knowledge check to finish this lesson.