# Quiz — Session 9 🟢 “Info or misinformation? Critical thinking in the age of AI »

**Program:** Applied AI — Beginner Level · **Instructor:** Yann Isola
**10 questions · Only one correct answer per question · Answer key commented at the end**

---

### Question 1
Why can generative AI give false information with great confidence?

- A. Because she is programmed to lie
- B. Because it predicts the most plausible sequence, without checking if it is true
- C. Because she is tired after too many questions
- D. Because she translates poorly from English

### Question 2
What do we call the phenomenon where an AI invents false but credible information?

- A. A bug
- B. A virus
- C. A hallucination
- D. An update

### Question 3
What is the difference between an **error** and **misinformation**?

- A. There are none
- B. The error is involuntary, the disinformation seeks to deliberately deceive
- C. The error is always more serious
- D. Misinformation only concerns videos

### Question 4
In the S.T.O.P. method, what does **P** mean?

- A. Share — disseminate the information as quickly as possible
- B. Prove — require written evidence
- C. Pause — notice the strong emotion and wait before sharing
- D. Police — report immediately to authorities

### Question 5
A huge “scoop” is not picked up by **any** known media. What should we conclude from this?

- A. This is proof that the media are hiding the truth
- B. This is a warning signal: such important information would normally be repeated everywhere
- C. That doesn’t mean anything
- D. It must be shared so that the media discover it

### Question 6
What tool allows you to find the origin of a suspicious image?

- A. Reverse image search
- B. The spelling checker
- C. Airplane mode
- D. Phone zoom

### Question 7
Why is the list of deepfake “visual clues” (weird hands, strange reflections, etc.) not sufficient protection?

- A. Because no one looks at the hands
- B. Because AI models are improving and correcting these defects: the indices age quickly
- C. Because deepfakes no longer exist
- D. Because the clues only work on black and white photos

### Question 8
A video shows a celebrity promising to “multiply your money” if you pay into a kitty. What is the RIGHT reflex?

- A. Pay a small amount to test
- B. Share the video to ask your friends for their opinion
- C. Do not contribute anything, check the source, and report: this is the typical profile of a deepfake scam
- D. Pay only if the video has a lot of views

### Question 9
What is the “liar’s dividend”?

- A. Money earned by the creators of fake videos
- B. The fact that, if everything can be false, liars can also deny the real evidence
- C. A tax on social networks
- D. The reward given to fact-checkers

### Question 10
A friend shares a poison in the class group. What is the best attitude?

- A. Make fun of him publicly so that everyone understands
- B. Say nothing, that’s not your problem
- C. Share the misinformation with others to ask for their opinion
- D. Inform him kindly (rather privately) by showing a reliable source

---
---

## ✅ Corrected commented

**1 — B.** The language AI predicts the most likely next word (seen in Session 3). It produces *plausible*, not *verified*.She does not “lie”: lying requires an intention, which she does not have.

**2 — C.** We speak of **hallucination**: an invented answer but presented with confidence. Typical examples: false book references, false quotes, invented biographies.

**3 — B.** It's all in the **intention**. To err is human and involuntary; disinformation is manufactured to deceive. Warning: a widely shared error can do as much damage as a misinformation.

**4 — C.** **P = Pause.** Misinformation is designed to trigger a strong emotion (anger, fear, indignation) that short-circuits thinking. Feeling this emotion is precisely the signal that you should wait 30 seconds before sharing.

**5 — B.** This is the reflex **T (Traces)**: to cross again. A major event is always covered by several independent media. “No one is talking about it except this account” = red flag, not proof of conspiracy.

**6 — A.** **Reverse image search** (Google Images, TinEye, Lens) shows where and when an image has already been published — very effective in unmasking an old recycled photo in a false context.

**7 — B.** Yesterday's defects (hands with six fingers, absent blinks) are corrected by today's models. The lasting lesson is not the list of clues, it is the **reflex to check the source**. Your eyes will betray you one day; your method, no. ⚠

**8 — C.** Promise of easy money + urgency + request for payment = scam, regardless of the quality of the video. A video that “looks real” is worthless as evidence when you are asked for money.

**9 — B.** If the public believes that "everything can be faked", then a person who is actually filmed doing something wrong can say "that's a deepfake!" ". This is why critical thinking is NOT “rejecting everything” but “verifying methodically”.

**10 — D.** We correct the **information**, not the **person**. A private, caring message with a reliable source is much more likely to work than public humiliation — and preserves the friendship.

---

**Indicative scale:** 8-10 ✅ excellent information detective · 5-7 🙂 good reflexes, review S.T.O.P. · <5 ⚠ reread the S.T.O.P. sheet. and repeat exercise 2.

*Quiz — Applied AI, Session 9 🟢 — Yann Isola.*