Artificial Intelligence: Principles & Techniques
A map of AI (search, logic, learning, uncertainty) and how the pieces fit.
AI tutor is turned off for this class
Use the CS221 lectures, notes, and assignments below to keep learning.
Lesson overview
A map of AI (search, logic, learning, uncertainty) and how the pieces fit.
Teacher script · 45 min
- 0–5
Hook
When a computer plays checkers or suggests a route home, it isn't thinking like you. It is trying lots of possibilities very fast and keeping the ones that score well. The method: generate options, score them, pick the best.
- 5–15
Explore
Students do the activity in pairs: Solve a maze twice: once exploring blindly, once always stepping toward the goal. Count the steps. The smarter rule wins: that's a 'heuristic.'
- 15–30
Explain
AI has a small toolbox. SEARCH explores possible futures (chess). CONSTRAINTS rule out impossible combinations (Sudoku, scheduling). LEARNING finds patterns in data (spam filters). PROBABILITY reasons when you can't be sure (medical diagnosis). Most real systems bolt several of these together.
- 30–40
Connect to the summit
Read aloud the "From the summit" section below. Ask students which ideas from today they recognize in it.
- 40–45
Check
Run the formative check below. Anyone who can explain a key term in their own words has it.
Student activity
Solve a maze twice: once exploring blindly, once always stepping toward the goal. Count the steps. The smarter rule wins: that's a 'heuristic.'
Slides
Formative check
- 1.In your own words, what is "State space"? (Looking for: Every situation the problem could possibly be in, and the moves between them.)
- 2.In your own words, what is "Heuristic"? (Looking for: A cheap guess at how close you are to the goal, used to search smarter.)
- 3.In your own words, what is "Markov Decision Process"? (Looking for: A formal way to describe choosing actions over time when outcomes are uncertain.)
Carry-away concepts
- State space
- Every situation the problem could possibly be in, and the moves between them.
- Heuristic
- A cheap guess at how close you are to the goal, used to search smarter.
- Markov Decision Process
- A formal way to describe choosing actions over time when outcomes are uncertain.
- Modeling
- Turning a vague real problem into a precise math problem you can actually solve.
From the summit · the Stanford source
Graduate survey: you implement search, constraint solvers, Markov decision processes, game-playing agents, and Bayesian networks, and argue about which tool fits which problem.
This module is based on a Stanford course.