VERSION A · LYCÉE GÉNÉRAL · Première: B1+ towards B2 · Terminale: B2 (LVA) · documents reaching towards C1
Session 3: AI gone wrong
Part 3. How can AI spread lies, and how can citizens protect themselves?
Objectives
- Understand chatbot errors, deepfakes and disinformation
- Use critical thinking to fact-check
- Make a complaint and write a formal apology
Warm-up
Students see the image of Pope Francis in a white puffer jacket that went viral in March 2023. Real or fake? It was generated with the AI tool Midjourney.
Documents
Document 5: The Air Canada chatbot. In 2024, a Canadian tribunal ordered Air Canada to pay a customer after its website chatbot gave him wrong information about a discount on fares. The airline argued that the chatbot was responsible for its own words; the tribunal rejected this argument.
Document 6: Disinformation. Key words: fake news, clickbait, echo chamber, filter bubble, deepfake, propaganda, phishing, scam, fact-check, censorship, privacy. Algorithms show us content we already like, which can lock us in a filter bubble.
Activities
- Vocabulary matching: ten words of Document 6 with their definitions.
- Fact-check method in four steps: Who wrote it? When? What do other sources say? Can I find the original?
- Suffix stress: propaGANda, informAtion, criticism, journalism.
- Intermediary task (interaction): a customer phones a tech company because its chatbot gave wrong advice about a refund.
- Intermediary task (written, summative): as the company, write an official apology to the customer (150 words).
Sentence starters
- I am calling because your chatbot told me...
- I would like you to...
- We sincerely apologise for...
- We take full responsibility for...
- In the future, we will...
Role play
Pairs: Student A is the customer, Student B is the customer service manager. The chatbot promised a refund that does not exist. A explains calmly and firmly; B listens, apologises and offers a solution.
Gap-fill worksheet
Word bank: deepfake, clickbait, echo chamber, fact-check, phishing, filter bubble, propaganda, privacy
- A ______ is a fake video or picture made with AI.
- A shocking title written to make you click is ______.
- An ______ is a space where you only hear opinions like yours.
- Before sharing, always ______ the information.
- A fake email asking for your password is ______.
- Algorithms can lock us in a ______.
- Messages spread by a state to influence people are ______.
- Protect your ______: do not share your personal data.
Answer key
- deepfake
- clickbait
- echo chamber
- fact-check
- phishing
- filter bubble
- propaganda
- privacy
Homework
Write the official apology letter (150 words): facts in the past tense, apology expressions, promises with will.
Assessment grid
| Criteria | Level 1 | Level 2 | Level 3 | Level 4 |
|---|---|---|---|---|
| Apology structure | Missing parts | Apology only | Facts, apology, solution | Complete, formal and convincing |
| Tenses (past, will) | Often wrong | Some errors | Mostly correct | Correct and varied |
| Register | Informal | Partly formal | Formal | Formal and polite throughout |
| Vocabulary | Basic | Some topic words | Varied | Rich and precise |