Decision Making Under Uncertainty
How to choose well when you can't see the future: the math behind self-driving cars and aircraft collision avoidance.
AI tutor is turned off for this class
Use the CS238 lectures, notes, and assignments below to keep learning.
Big Idea
Representation & Reasoning
Grade bands
K-2 · 3-5 · 6-8 · 9-12
AI literacy pillar
How AI works · Ethics
Lesson overview
How to choose well when you can't see the future: the math behind self-driving cars and aircraft collision avoidance. This module climbs from an everyday intuition to the real mechanism, then names the Stanford course it descends from.
Teacher script · ~45 min
- 0–5
Hook
Should you bring an umbrella? You don't know if it'll rain; you only know a chance. Good decisions aren't about being right every time; they're about being right on average given what you know. That single shift is the whole field.
- 5–15
Explore
Students do the activity in pairs: Use it for real: should you study or sleep before a test you're 70% ready for? Put numbers on it and compute.
- 15–30
Explain
Real life isn't one choice; it's a sequence, where each move changes what comes next, and sometimes you can't even see the full situation (fog, sensor noise). The trick is to reason backward from good futures to today's best move, and to value information itself, because knowing more changes what you should do.
- 30–40
Connect to the summit
Show students this is the real thing professionals build: CS238 (AA228), the real thing. How to choose well when you can't see the future: the math behind self-driving cars and aircraft collision avoidance.
- 40–45
Check
Run the formative check below. Anyone who can explain a key term in their own words has it.
Student activity
Use it for real: should you study or sleep before a test you're 70% ready for? Put numbers on it and compute.
Slides
Formative check
- 1.In your own words, what is "Expected value"? (Looking for: The average payoff of a choice, weighing each outcome by how likely it is.)
- 2.In your own words, what is "Utility"? (Looking for: A number capturing how much you actually value an outcome, not always the dollar amount.)
- 3.In your own words, what is "POMDP"? (Looking for: A decision problem where you can't directly see the true state, only noisy hints of it.)
Carry-away concepts
- Expected value
- The average payoff of a choice, weighing each outcome by how likely it is.
- Utility
- A number capturing how much you actually value an outcome, not always the dollar amount.
- POMDP
- A decision problem where you can't directly see the true state, only noisy hints of it.
- Value of information
- How much a decision improves if you go gather one more fact before choosing.
From the summit · the Stanford source
You formalize sequential decisions as POMDPs, solve them with value iteration and online planning, and handle cases where you can't even observe the true state of the world.
This module descends from CS238 at Stanford. Students who climb the full ladder arrive here.