Principles of AI ›  Unit 1

Unit 1 Inside the Machine

Build a language model from dice, train a classifier on your own data, then learn to break chatbots on purpose. Ends with a Red-Team Report on a system you have to attack and defend.

🕐 ~4 weeks🅰️ 6 Activities🅿️ 2 Projects📅 16 class periods
Principles of AI Home

The Path Through Unit 1

A local print shop owner just paid for an AI tool a salesperson swore would "run the front office." Now she wants to know what it should and shouldn't touch. Read the full case →
AActivity6 days total
1-1-1Next-Word Dice — Build a Tiny Language Model
2 days→
1-1-2Train, Test, Trust — Teachable Machine & the Confusion Matrix
2 days→
1-1-3Matchbox Nim — A Learner Made of Cups and Beads
2 days→
PProject🕐 3 class days

1-1-4  Classifier & Model Card

Pick a real need, collect your own data, and train a classifier. Then test it on images it has never seen and report honestly where it works and where it breaks.

Turn in: Trained classifier + model card with accuracy and failure casesOpen →
AActivity4 days total
1-2-1Password Wars — Prompt Injection & Guardrails
2 days→
1-2-2Stump the Bot — Catching Confident Errors
1 day→
1-2-3Real or Fake? — Synthetic Media & Scams
1 day→
PProject🕐 3 class days

1-2-4  Red-Team Report

Attack a chatbot setup the way a security tester would, then defend it. Document every failure you find and recommend fixes someone could actually ship.

Turn in: Red-team report: attacks tried, failures logged, fixes recommendedOpen →