Français

Slides — Session 1 🟢

“What is artificial intelligence?”

Program: Applied AI — Beginner Level · Instructor: Yann Isola
24 slides · presenter notes included
(Palette: ink #1A2230, teal #0F7A6C, copper #B4612A, light teal #E9F6F3, background #F4F7F6)

Slide 1 — Title

What is artificial intelligence? 🤖

Applied AI · Beginner Level · Session 1

Yann Isola

Slide 2 — Opening question

Who has used AI today? 🙋

Slide 3 — What is AI for you?

🧠 Icebreaker

An image, a word, a fear, a desire…

Slide 4 — The definition (THE slide to remember)

AI is a program that learns from examples

…instead of being programmed rule by rule.

AI = Artificial Intelligence

Slide 5 — The bicycle analogy

Nobody learns to ride a bike from a manual 🚲

  • We try 🚴
  • We fall 😅
  • We adjust 🔧
  • We succeed 🎉

Slide 6 — The great adventure of AI

70 years of history in 5 moments 🕰️

1950 · 1966 · 1997 · 2016 · 2022

Slide 7 — 1950: Alan Turing

“Can machines think?” 🇬🇧

  • World War II codebreaker (he helped break Enigma ciphers)
  • 1950: he proposed the Turing test — if you cannot tell whether you are talking to a person or a machine, the machine passes
  • At the time, a computer was less powerful than a school calculator!

Slide 8 — 1966: ELIZA, first chatbot

The virtual therapist that trapped everyone 💬

  • ELIZA only reformulates your sentences
  • A very simple trick… and yet people confided in it for hours
  • Its creator himself was frightened by it

Slide 9 — 1997: Deep Blue vs. Kasparov

The day the machine beat the champion ♟️

  • Garry Kasparov: the best chess player in history
  • Deep Blue (IBM): ~200 million positions calculated per second ⚠
  • Brute force, not learning: it calculated; it did not “think”

Slide 10 — 2016: AlphaGo and move 37

The move experts mistook for a bug 🀄

  • Go: more possible positions than atoms in the observable universe ⚠
  • Impossible to calculate everything → AlphaGo learned: millions of games, then games against itself
  • Match 2, move 37: a move so strange that the commentators thought it was an error... it was a stroke of genius

Slide 11 — 2022: ChatGPT

AI reaches everyone’s pocket 🚀

  • 100 million users in 2 months ⚠ — adoption record at the time
  • GPT = “Generative Pre-trained Transformer” — remember: trained on huge amounts of text to predict next words
  • For the first time, anyone can chat with an AI

Slide 12 — The 3 types of AI

Narrow · General · Super 🎯🧠🌌

Exists? What is it?
Narrow AI 🎯 ✅ everywhere One talent at a time
General AI 🧠 ❌ science fiction Versatile like a human
Superintelligence 🌌 ❌ philosophical debate Surpasses humans in everything

Slide 13 — The swimming champion analogy

An Olympic champion… who doesn’t know how to walk 🏊

  • AlphaGo crushes humans in go…
  • …but doesn't know how to play tic-tac-toe, or say hello, or add up
  • Each AI = only one talent

Slide 14 — PAUSE

☕ Break — 10 minutes

Meanwhile: the interactive timeline is freely accessible!

Slide 15 — How does it work?

Data → Learning → Prediction

The 3 stages of ALL artificial intelligence

Slide 16 — The story of the baby and the cats

How YOU learned to recognize a cat 🐱

  1. Data: we showed you lots of cats
  2. Learning: your brain found the pattern on its own (ears, whiskers, etc.)
  3. Prediction: you recognize a cat you have NEVER SEEN before

Slide 17 — You are the AI (demo)

It's up to you: blorp or not blorp? 👾

Slide 18 — 3 truths about machine learning

What to remember 📌

  • 📦 No data, no AI — data is the fuel
  • 🔍 AI finds patterns, it doesn’t “understand” like us
  • 🎲 A prediction can be wrong. Always.

Slide 19 — Game: AI or not AI?

🎮 AI or not AI?

Netflix? The calculator? The supermarket door? The photo filter?

The magic clue: did it learn from examples? could it be wrong?

Slide 20 — Main activity

🤖 Your first conversation with an AI

4 missions: Get to know it · Ask it to amaze you · Try to catch it out · The reverse Turing test

⚠️ Golden rule: no personal information in the chatbot!

Slide 21 — Word of the day: hallucination

When AI invents… with total confidence 🎩

  • AI produces plausible text, not necessarily true
  • It can sound just as confident when it is wrong as when it is right
  • 🥇 Golden rule: check AI output; never take it at face value

Slide 22 — Myths vs reality

3 myths to debunk 🔨

Myth Reality
“AI thinks” 🧠 It calculates probabilities. The parrot that says “hello” doesn’t wish you a good day
“AI will replace everyone” 😱 Nuanced: it replaces tasks, transforms jobs, creates new ones
“AI is infallible” 🎯 It is often wrong — and with confidence!

Slide 23 — AI jobs

What if it was YOUR future job? 💼

  • 🔍 Data scientist — the data detective
  • 🏗️ ML Engineer (Machine Learning) — they build and train models
  • 🗣️ Prompt engineer — designs effective instructions and workflows for AI systems
  • ⚖️ AI ethicist — asks, “Is this fair? Who is responsible?”

AI is not just for math people: we need creative people, writers and humanities specialists, skeptics!

Slide 24 — Conclusion

What you know now 🎓

  • ✅ AI = a program that learns from examples
  • ✅ 70 years of history: Turing → ELIZA → Deep Blue → AlphaGo → ChatGPT
  • ✅ Data → learning → prediction
  • ✅ AI is everywhere… but not in the supermarket door 😄
  • ✅ It does not think, and it is not infallible — we always check

🔜 Next session: we open the hood — how does a machine really learn?

Notes: welcome, smile, announce the promise: “In 2 hours, AI will no longer be magic for you. And it's even better that way.”

Notes: raise your hands. Few hands. Then: “Who used a GPS? Autocorrect? Watched a recommended video?” → all hands. First wow moment: you've been using AI for years without knowing it.

Notes: table tour or post-its. DO NOT correct ANYTHING. Write everything down on the board — we’ll come back to it at the end.

Notes: have the definition repeated out loud. This is THE sentence of the session.

Notes: “AI learns the same: from examples and mistakes. Not by the manual.” Introduce ML = Machine Learning = machine learning.

Notes: announce: “5 stories, not 5 dates.” Switch to the interactive timeline of the web page if projector + browser available.

Notes: tell, not recite. Turing = the grandfather of this whole story.

Notes: Key lesson: Humans easily attribute intelligence to what speaks. That hasn’t changed in 2022…

Notes: IBM = International Business Machines. Emphasize: Deep Blue was not learning. The real revolution comes later.

Notes: wow moment. “AI surprised its own creators.” AlphaGo = DeepMind (Google).

Notes: ask who has already used it. Transition: “But then, how does it work in there?”

Notes: “All the AI you will encounter this year is narrow.” Anticipate the ChatGPT question (see FAQ in the guide).

Notes: this image sticks in people's minds. That's the goal.

Notes: leave the web page open on a computer. The curious will come and play — that’s intentional.

Notes: announce: “We are going to experience it with a cat story.”

Notes: “No one gave you the definition of cat. AI is exactly the same — in silicon.”

Notes: THE participatory demo. Show the 4 examples, have people guess the 5th. Then: “How did you know? Nobody gave you the rule! You have just done machine learning.” The rules found without being written = everything is there.

Notes: what if the examples are bad? → AI learns from errors and biases. Seed planted for the myths slide.

Notes: switch to the interactive game on the web page. Vote in teams or by show of hands. Emphasize the pitfalls: automatic door (sensor, NOT AI), calculator (fixed rules, NOT AI).

Notes: pairs, 20 min, exercise sheet 3. Circulate between groups. Collect the best “hallucinations” for the debrief.

Notes: ask the pairs for their best findings from mission 3. Laughing together at the hallucinations = best vaccination against gullibility.

Notes: on myth 2, be honest: it's a real social issue, certain jobs change quickly. The best defense: understand the tool. That's exactly what they're doing today.

Notes: emphasize the diversity of profiles. AI ethics is recruiting philosophers and lawyers.

Notes: return to the icebreaker board: “What are we keeping? What are we crossing out?” Give the take-home challenge: note 5 times you encounter AI during the week. End with: “AI is neither magical nor monstrous. It is a powerful, imperfect, fascinating tool. And now you know how it learns.”