# Quiz — Session 10 🟢
## Overall review of the beginner course (Sessions 1 → 9)

**Program:** Applied AI — Beginner Level · Instructor: Yann Isola
**Format:** 10 multiple choice questions · only one correct answer per question · answer key commented at the end of the document
**Recommended use:** in teams, answers on the slate (A/B/C/D), correction on the fly

---

## Questions

### Question 1 — The definition of AI (S1)
Which phrase best describes artificial intelligence?

- **A.** A robot that thinks exactly like a human
- **B.** A program that learns from examples instead of just following rules written one by one
- **C.** A database that contains all possible answers
- **D.** A magical program that cannot be explained

### Question 2 — How AI learns (S2)
What is the correct order of an AI's learning cycle?

- **A.** Prediction → data → training
- **B.** Training → prediction → data
- **C.** Data → training → prediction
- **D.** Rules → magic → result

### Question 3 — The language of machines (S3)
When a chatbot writes a sentence, what is it actually doing?

- **A.** He copies a sentence found on the internet
- **B.** It predicts, piece by piece, the most probable sequence of text
- **C.** He understands the world like a human before responding
- **D.** He chooses sentences at random from a dictionary

### Question 4 — Prompting (S4)
Which of these prompts is most likely to give a good result?

- **A.** “Write a text. »
- **B.** “Talk to me about stuff.” »
- **C.** “You are an SVT teacher. Explain photosynthesis to a 6th grade student in 5 simple sentences with an example. »
- **D.** “Answer quickly!!! »

### Question 5 — Multi-sense generative AI (S5)
Which of these tasks falls under **generative** AI?

- **A.** Sort emails into “spam” and “non-spam”
- **B.** Create an image from a written description
- **C.** Measure the temperature of a room
- **D.** Count the number of words in a text

### Question 6 — Ethics (S6)
A recruitment AI more often rejects female applications because it was trained on old files that were predominantly male. What is this problem called?

- **A.** A display bug
- **B.** A bias (inherited from training data)
- **C.** A hallucination
- **D.** An internet connection problem

### Question 7 — AI in real life (S7)
In professions (health, agriculture, transport, etc.), the most realistic role of AI today is:

- **A.** Completely replace professionals
- **B.** Assist humans, who retain decision-making and responsibility
- **C.** Decide alone, without any human control
- **D.** Of no use outside of video games

### Question 8 — Creativity (S8)
Which attitude best sums up co-creation with AI?

- **A.** “AI creates, I watch. »
- **B.** “I copy the first result of the AI as is and I sign it. »
- **C.** “The AI suggests ideas, I choose, I modify, I add my personal touch. »
- **D.** “Human creativity is no longer useful. »

### Question 9 — Critical thinking (S9)
You see a spectacular photo shared online. What is the RIGHT reflex?- **A.** Share it quickly before everyone else
- **B.** Believe her if she has a lot of “likes”
- **C.** Check the source and cross-check with other reliable media before believing or sharing it
- **D.** Only believe images that confirm what I already think

### Question 10 — Your project & your future (S10)
For an AI application project to be solid, it must be:

- **A.** Secret, complicated and impossible to explain
- **B.** Useful, realistic (AI capacity + identified data) and responsible (anticipated risks)
- **C.** Based on a magical AI that does absolutely everything
- **D.** Profitable above all, regardless of the risks

---

## ✅ Corrected commented

### Question 1 → **B**
The founding idea of the course: AI learns from **examples** (data) rather than being programmed rule by rule. Neither artificial human (A), nor simple stock of answers (C), and above all not magic (D) — it's engineering that we can understand!

### Question 2 → **C**
**Data → training → prediction**: we show examples, the model adjusts its settings, then it makes predictions on new things. Remember: without data, no AI.

### Question 3 → **B**
A language model calculates the **most likely sequence**, token by token. He does not copy (A) and does not “understand” like us (C) — hence the possible hallucinations, and the importance of checking (Q9!).

### Question 4 → **C**
The recipe for a good prompt: **context + specific task + expected format** (+ target audience). The clearer the prompt, the better the response. “Answer quickly!!! » has never improved a result 😄.

### Question 5 → **B**
Generative = which **creates new content** (text, image, sound, video). Sorting emails (A) is classification: it's AI, but not generative. C and D aren't even AI.

### Question 6 → **B**
This is a **bias**: the AI reproduces — and sometimes amplifies — the imbalances present in its training data. Lesson from Session 6: Biased data makes for biased AI, hence the importance of checking and correcting.

### Question 7 → **B**
The model that works: the AI **assists** (detects, suggests, pre-sorts), the human **decides** and remains responsible. The doctor validates the diagnosis, the farmer decides on the treatment.

### Question 8 → **C**
Co-creation: AI is a **partner of ideas**, not a distributor of finished works. Choice, taste and intention remain human. (And signing a raw result as is, answer B, also raises a question of honesty!)

### Question 9 → **C**
The triple reflex of session 9: **stop, check the source, cross**. Likes prove nothing (B), speed of sharing is the ally of fake news (A), and D describes confirmation bias — our worst enemy.

### Question 10 → **B**
The 3 golden rules of the final project: **useful** (real problem, real audience), **realistic** (what AI capacity? what data?), **responsible** (what risks? what solutions?). This is exactly the summary of the 9 sessions in one grid!

---

## 📊 Team Score

- **9-10 correct answers:** 🌳 Graduated with honors! The course is acquired.
- **7-8:** 🌿 Very solid — just reread the answer key for missed questions.
- **5-6:** 🌱 The basics are there; the final project will consolidate everything.
- **< 5 :** 💪 Don't panic — the commented correction is made for you, and the project will put everything back in place.

*Congratulations for making it this far. On to the project! 🚀*