Interactive Exercises - Learn by Doing
Knowledge Without Practice is Just Theory
You've learned the concepts. Now it's time to cement them through hands-on exercises, quizzes, and practical activities. Let's do this!
📝 Chapter 1 Quiz: AI Fundamentals
Self-Assessment Questions:
Q1: What is AI fundamentally?
Q2: Which AI type exists today?
Q3: How many times do you likely interact with AI daily?
Try This: Pattern Recognition Exercise
Materials: Pen and paper
Time: 5 minutes
- Write down 10 email subject lines from your spam folder
- Identify 3 common patterns (words, punctuation, style)
- Write a "rule" to detect spam
- Test your rule on 5 new emails
- Congratulations! You just simulated AI training
Reflection: How accurate was your rule? This is exactly how spam filters learn! They identify patterns in thousands of spam emails and create rules to detect them.
📝 Chapter 2 Quiz: How AI Learns
Knowledge Check:
Q1: AI learning is most similar to:
Q2: Why does AI need so much data?
Hands-On: Teach Your Brain "AI"
The 20 Questions AI Simulator
Setup: Play with a friend
- Friend thinks of an animal
- You ask yes/no questions (max 20)
- Track your questions in categories:
- Physical traits
- Habitat
- Behavior
- Size
Track Your Performance:
Round 1: ___ questions to guess
Round 2: ___ questions to guess
Round 3: ___ questions to guess
Learning moment: Notice how you narrow down possibilities? That's decision tree learning! Are you getting faster? You're learning patterns just like AI!
🏗️ Chapter 3 Exercise: Build a Mini Transformer
Word Relationship Map (Paper Version)
Materials: Paper, pen
Time: 15 minutes
Step 1: Pick a sentence
Step 2: Create relationship scores (1-10)
Word 1: doctor
→ treated: Score: 9 (strong action link)
→ patient: Score: 8 (common association)
→ hospital: Score: 9 (typical location)
→ the: Score: 2 (weak link)
Step 3: Your turn!
Pick your sentence: _______________
Word 1: _____
→ Word 2: _____ Score: __
→ Word 3: _____ Score: __
→ Word 4: _____ Score: __
🎓 Key Insight: High scores = strong relationships (like "doctor" and "hospital"). This is exactly how transformer attention weights work! You just built a mini transformer by hand!
🔍 Understanding Transformers: Attention Activity
Attention Mechanism Exercise
Materials: Highlighter, any article
Time: 10 minutes
- Read a news article paragraph
- Highlight the 3 most important words
- Now highlight words that relate to those 3
- Draw arrows showing relationships
Example:
"The cat sat on the mat because it was soft"
• cat → it (reference)
• mat → soft (description)
• sat → on (action-location)
🎉 Congratulations! You just performed attention mechanism manually! This is exactly how transformers understand which words in a sentence relate to each other.
⚖️ Model Size Comparison: Performance Testing Lab
Compare Models Yourself
Task: "Write a recipe for chocolate cake"
Test on 3 platforms and compare results:
1. Small Model (Bing Chat/Bard)
Time: _____ seconds
Quality (1-10): _____
Details: _________________
2. Medium Model (ChatGPT-3.5)
Time: _____ seconds
Quality (1-10): _____
Details: _________________
3. Large Model (GPT-4/Claude)
Time: _____ seconds
Quality (1-10): _____
Details: _________________
📊 Observe: Larger models typically provide better quality and more detailed responses, but they're slower and more expensive. Find the sweet spot for your needs!
🔤 Chapter 5 Exercise: Tokenization
DIY Tokenizer: Breaking Down Text
Take this sentence: "Understanding tokenization is important"
1. Letter tokens (26 tokens):
U-n-d-e-r-s-t-a-n-d-i-n-g t-o-k-e-n-i-z-a-t-i-o-n...
2. Word tokens (4 tokens):
[Understanding] [tokenization] [is] [important]
3. Subword tokens (6-8 tokens):
[Under][stand][ing] [token][ization] [is] [import][ant]
Which is most efficient?
Answer: ___________
💡 Pro Tip: Subword tokenization balances efficiency and vocabulary size. It's what modern models like GPT use!
Token Counting Challenge
Estimate tokens (1 token ≈ 4 characters):
"Hello" = ___ tokens
Answer: 1 token
"Antidisestablishmentarianism" = ___ tokens
Answer: 6-8 tokens
"The quick brown fox jumps over the lazy dog" = ___ tokens
Answer: ~10 tokens
This entire paragraph = ___ tokens
Answer: ~30 tokens
🎯 Understanding tokens: This matters because APIs charge per token, and models have token limits. Knowing how to estimate saves money and prevents errors!
🧠 Chapter 6: Neural Network Simulator
Human Neural Network (Group Activity)
Setup: 3+ people needed
Task: Recognizing shapes
Assign Roles:
- Person 1 (Neuron 1): "Sees" edges - counts how many sides
- Person 2 (Neuron 2): "Sees" corners - counts how many angles
- Person 3 (Output Neuron): Makes final decision based on inputs
Process Example:
- Show a shape (triangle, square, circle)
- Person 1: "I see 3 edges!" (or 4, or curved)
- Person 2: "I see 3 corners!" (or 4, or none)
- Person 3: Combines info → "It's a triangle!"
Track Your Results:
🎓 Debrief: Each "neuron" (person) has a specific job! Person 1 detects one feature, Person 2 detects another, and Person 3 combines them. This is EXACTLY how neural networks work - specialized neurons detecting specific features, then combining them for the final answer!
📚 Build Your First Dataset
Project: Email Classifier
Goal: Create 20 training examples
Template:
Example #___
Input: [Email subject line]
Category: [Work/Personal/Spam]
Why: [Your reasoning]
Categories to Include:
Quality Check:
💡 Pro Tip: The quality of your dataset determines the quality of your AI. Take time to create diverse, accurate examples!
🔄 Training Simulation: Watch AI Learn
Train an "AI" to Recognize Fruits
Round 1: Initial Guesses (No Training)
• Apple → AI guesses: "Ball?" ❌
• Banana → AI guesses: "Stick?" ❌
• Orange → AI guesses: "Ball?" ❌
• Grape → AI guesses: "Marble?" ❌
• Strawberry → AI guesses: "Heart?" ❌
Accuracy: 0% 😢
Adjustment Phase:
AI realizes shape alone isn't enough. Adjusts to focus on color + shape!
Round 2: After Training
• Apple → AI guesses: "Red fruit?" ✓
• Banana → AI guesses: "Yellow fruit?" ✓
• Orange → AI guesses: "Orange fruit?" ✓
• Grape → AI guesses: "Purple fruit?" ✓
• Strawberry → AI guesses: "Red berry?" ✓
Accuracy: 100% 🎉
Track Your Understanding:
Round 1 Accuracy: ____% (before learning)
Round 2 Accuracy: ____% (after learning)
Round 3 Accuracy: ____% (refinement)
🎨 Chapter 9: Fine-tuning Project
Personalize Your AI: Create Your Writing Style
Step 1: Collect Writing Samples
Step 2: Identify Your Patterns
Common phrases you use:
Average sentence length:
Your tone (formal/casual/humorous):
Step 3: Create Training Format
Input: "Write about [topic]"
Output: "[Your style writing example]"
// Repeat for 10-20 examples
Step 4: Test with AI
- Give AI your writing examples
- Ask it to mimic your style on a new topic
- Compare AI output to your original writing
- Rate similarity (1-10): _____
Evaluation Criteria:
🎯 Project Goal: This exercise demonstrates fine-tuning! You're teaching AI your specific style - the same way companies fine-tune models for customer service, legal writing, or medical documentation.
🔒 Chapter 10: Local vs Cloud Comparison
Privacy Audit: Track Your Data
Track for One Day:
Cloud AI Uses (Data Sent):
Could This Be Local Instead?
Calculate Your Privacy Score:
Data sent to cloud: ___ times today
Could have been local: ___ times
Privacy improvement potential: ___%
Decision Matrix: When to Use What
| Task Type | Local | Cloud |
|---|---|---|
| Private docs | ✓ Best | ✗ Risky |
| Complex tasks | △ Limited | ✓ Best |
| No internet | ✓ Works | ✗ Fails |
| Quick drafts | ✓ Fast | ✓ Fast |
🎯 Key Insight: Most people send 20-50 AI requests to the cloud daily. By identifying which could run locally, you can protect sensitive data while still getting AI assistance. Privacy isn't all-or-nothing - it's about smart choices!
📈 Your Learning Progress Tracker
Week 1: Foundation
Week 2: Hands-On
Week 3: Building
Week 4: Advancing
🗓️ 30-Day AI Mastery Challenge
Commit to one action per day for 30 days and transform from AI beginner to practitioner:
WEEK 1: UNDERSTAND
WEEK 2: EXPERIMENT
WEEK 3: BUILD
WEEK 4: MASTER
🚀 Project Templates: Your AI Journey
Beginner Project: AI Journal
Perfect first project: Use AI to enhance daily journaling and track learning
Week 1: Setup
Week 2: Daily Practice
Week 3: Analysis
Week 4: Share
Template Prompts: "Help me reflect on...", "What patterns do you see in...", "Summarize my learning this week about..."
Intermediate Project: Content Assistant
Build a specialized AI assistant for your specific content needs
Planning Phase:
Define content type:
Identify your patterns:
Building Phase:
Choose model for your use case:
Refinement Phase:
Success Metrics: 80% reduction in drafting time, consistent tone across all content, 90% satisfaction rate
Advanced Project: Domain Expert
Create a specialized AI for your field (legal, medical, technical, etc.)
Research Phase (Week 1):
Development Phase (Week 2-3):
Deployment Phase (Week 4):
Success Criteria: 95% accuracy on domain tasks, positive feedback from experts, reduces work time by 50%+
🤔 Reflection Questions
After each chapter, take a moment to reflect:
1. What surprised me most?
2. What can I apply today?
3. What do I want to explore more?
4. Who can I share this with?
5. What's my next action?
Key Takeaways
- ✓Practice makes perfect - reading alone won't make you an AI expert
- ✓Quizzes test understanding - Chapter 1 & 2 quizzes validate your foundational knowledge
- ✓Hands-on exercises cement concepts - from transformers to tokenization to neural networks
- ✓Building datasets teaches AI thinking - quality data determines quality results
- ✓Fine-tuning creates personalized AI - learn your writing style and domain expertise
- ✓Privacy audits reveal choices - balance local vs cloud based on your needs
- ✓Project templates guide your journey - from beginner journal to advanced domain expert
- ✓The 30-day challenge creates habits - consistency beats intensity every time
- ✓Reflection deepens learning - always ask yourself what you learned
"Learning by doing is 10x more effective than reading alone. Complete these exercises to truly master AI."
Ready for Real-World Inspiration?
You've practiced the concepts. Now see how real people and companies are using AI to transform their work and lives!
Chapter 18: Real-World Case Studies →