Day 1: Welcome to the Future Intelligence Lab!

Simon Guest

Hello, I’m Simon!

What is Artificial Intelligence (AI)?

Traditional software: you write the rules → computer follows them
AI: Software that learns from examples, rather than following rules someone wrote
The definition keeps shifting as AI improves — that’s part of what makes it interesting

Why is AI Popular Right Now?

Three things came together at the same time:

Data — the internet created training data at a scale that never existed before
Compute — GPUs made it feasible to train on that data
Algorithms — breakthroughs in neural network design unlocked new capabilities

What AI is NOT

Not magic — it’s math and statistics applied to data
Not conscious — it has no understanding, only pattern matching
Not always right — it fails in predictable and unpredictable ways
Not one thing — it’s a family of very different techniques

Does It Use AI?

❌ No 🤔 Maybe ✅ Yes

ChatGPT: Does It Use AI?

❌ No 🤔 Maybe ✅ Yes

ChatGPT: Does It Use AI?

✅ Yes

Fitness Tracker: Does It Use AI?

❌ No 🤔 Maybe ✅ Yes

Fitness Tracker: Does It Use AI?

✅ Yes

Calculator: Does It Use AI?

❌ No 🤔 Maybe ✅ Yes

Calculator: Does It Use AI?

❌ No

Washing Machine: Does It Use AI?

❌ No 🤔 Maybe ✅ Yes

Washing Machine: Does It Use AI?

🤔 Maybe but most often ❌ No

Music App: Does It Use AI?

❌ No 🤔 Maybe ✅ Yes

Music App: Does It Use AI?

✅ Yes

Smartphone Camera: Does It Use AI?

❌ No 🤔 Maybe ✅ Yes

Smartphone Camera: Does It Use AI?

✅ Yes

Video Game: Does It Use AI?

❌ No 🤔 Maybe ✅ Yes

Video Game: Does It Use AI?

🤔 Maybe and often ✅ Yes

AI Around You

Think of other things that you use everyday.

Put them in the NO/MAYBE/YES categories in this board:

Join Board

Types of AI

There are many different types of AI
Three main types over the next two weeks
- Supervised ML
- Computer Vision
- Generative AI

Supervised ML (Machine Learning)

Traditional programming
- Write rules → computer follows them
Machine learning
- Show examples → computer learns the rules
What does Supervised mean?
- We label the data; the AI model discovers finds the pattern

Supervised ML (Machine Learning)

Two types of Supervised ML
- Regression
- Classification

Regression: Predicting a Number

10 → 20 → 30 → 40 → ?

Regression: Predicting a Number

1 → 2 → 1 → 3 → ?

Regression: Predicting a Number

7 → 2 → 0 → 5 → ?

Regression: Predicting a Number

The model outputs a continuous value
Predict tomorrow’s temperature from today’s weather
Predict house price from size and location
Predict livestock mortality from drought and temperature
We’ll build one of these on Day 3 with real Mongolian data

Classification: Predicting a Category

5°C → 🧥

18°C → 👔

30°C → 👕

12°C → ?

Classification: Predicting a Category

The model outputs a label from a fixed set
Spam or not spam?
Cat, dog, or bird?
Walking, still, or waving?
We’ll build a gesture classifier on Days 4–5 using a micro:bit

Computer Vision: How Computers See

A digital image is just a grid of numbers (pixels)
Each pixel has a value: 0 (black) → 255 (white)
A 640×480 photo = 307,200 numbers
The computer processes grids of numbers — it has never truly “seen” anything

Computer Vision All Around You

Face unlock on your phone
Photo search (“find all photos with a dog”)
Self-driving cars detecting pedestrians
Medical imaging: spotting tumors in X-rays
We’ll build our own CV applications on Day 6

Generative AI: Creating, Not Just Predicting

Supervised ML predicts from existing data
Generative AI creates new content
- Text, images, audio, video, code
Trained on vast amounts of human-created work
Two main types: LLMs and Diffusion Models

Large Language Models (LLMs)

Trained to predict: what word comes next?
Do this billions of times on internet-scale text
Emergent result: answer questions, write code, explain concepts
Examples: ChatGPT, Claude, Gemini
“The capital of Mongolia is ___”

Diffusion Models: From Noise to Image

Start with random noise (static)
Repeatedly denoise, guided by a text description
After ~50 steps: a photorealistic image
Examples: DALL·E, Midjourney, Stable Diffusion
We’ll generate images on Day 9

Your Two-Week AI Journey

Days	Topic	What You’ll Build
Days 2–3	Supervised ML	Data analysis & regression
Days 4–5	Supervised ML	Gesture classifier on micro:bit
Days 6–7	Computer Vision	CV mini-project
Days 8–9	Generative AI	Chatbot + image generation
Day 10	Capstone	Your own AI project

“How are we going to create all of this in Python?”

Introducing Notebooks

What is a Notebook?

An interactive document that combines:
- Python code that can be executed
- Rich text explanations (markdown)
- Visualizations and outputs
Think of it as a computational story
- Tell a story with code, data, and explanations
Originally designed for data science and research
Now, a fundamental tool for working with AI

Anatomy of a Python Notebook

Cells: Building blocks of notebooks
- Code cells: Executable Python code
- Markdown cells: Text, headings, images, equations
Outputs: Results appear directly below code cells
- Text, tables, plots, interactive widgets
Kernel: The computational engine running your code
- Maintains state between cell executions

Introducing Jupyter-K12

Uses the Jupyter document format
Notebook platform designed for students
Extra features for in-classroom use
Created by me :)
You are the first students in the world to use this!

Demo

“Hello World” Notebook on Jupyter-K12

Hands-On

Open the Curriculum Page

https://simonguest.github.io/codercub

Find the “Hello World” Notebook in Day 1

Click on this button: Open In Jupyter K12

Run all of the code in the notebook

Demo

Configuring the OPENROUTER_API_KEY

Hands-On

Go into Settings

Scan QR Code

Make sure that OPENROUTER_API_KEY is set

Demo

“AI in Action” Notebook on Jupyter-K12

Hands-On

Open the “AI in Action” Notebook

Run through all of the cells!

Discussion

What worked? What didn’t work?

Tomorrow

Use a lot of data in Python!
pandas and matplotlib
Using weather data from Ulaanbaatar