Thinking Fast and Slow × AI
Course index · 10 lessons · Kahneman's dual-process ideas applied to how people use and judge AI output
Mission
Improve my own AI use by noticing when fast thinking accepts or dismisses AI output without checking it, and by building a short check to run before acting on it. Starting point: new to the material. See MISSION.md .
Lessons
01 System 1 and System 2 in AI use The two modes of thought, automation bias, and why fluent output feels right.
02 Anchoring: the first number wins How a number in your prompt steers the answer, and how an answer steers you.
03 Availability: what comes to mind Why one vivid AI failure can distort your sense of how reliable it is.
04 Representativeness and base rates When a judgment fits a stereotype and ignores how common the outcome is.
05 Framing: same facts, different answers How wording changes AI verdicts, and the opposite-framing test.
06 Fluency feels true Why polished text, repetition, and citations lower your guard.
07 Confirmation bias meets sycophancy Leading questions get agreement, and how to ask for the strongest counterargument.
08 Overconfidence and calibration Why longer explanations raise confidence without raising accuracy.
09 Substitution and what you see Answers to easier questions, and answers built on missing information.
10 A pre-action check for AI output Capstone: one short check that combines the lessons before a decision.
Reference
Glossary . Definitions of every term used in the lessons, with links back to where each is taught.
Resources . The book, research, and preprints, ordered by trust. Full notes in RESOURCES.md .
Questions? Ask your teacher in this conversation about any lesson. Start with
Lesson 0001 .