Glossary

The vocabulary of thinking fast and slow with AI. Each term links to the lesson where it is taught.

No matching terms found.
A
Anchoring Bias
The pull an initial number or suggestion exerts on later estimates, even when it is irrelevant. In AI use it runs both ways: a number in your prompt steers the answer, and the answer steers you.
Automation bias Bias
Over-reliance on automated recommendations. Produces errors of commission (following a wrong recommendation) and errors of omission (missing what the system did not flag).
Availability heuristic Bias
Judging how common or likely something is by how easily examples come to mind. Recent or vivid cases feel more frequent than they are.
B
Base rate neglect Bias
Underweighting how common an outcome or category is when a description seems to fit a type. The engineer-lawyer study is the classic demonstration.
C
Calibration
The match between stated confidence and actual accuracy. A well-calibrated judge is right about 80 percent of the time when saying 80 percent.
Cognitive ease
The feeling that information is easy to process. Kahneman's term for why familiar, fluent statements feel true and safe.
Cognitive offloading
Using a tool, such as a calculator or calendar, to handle a routine sub-task while your own reasoning stays in charge. Contrast with cognitive surrender.
Cognitive surrender Bias
The unconscious habit of accepting AI outputs, automated recommendations, or outside judgments as your own without critical evaluation or mental struggle. Term from Shaw and Nave (2025); Improving uses it for this course. Unlike offloading, it hands over the core judgment itself.
Confirmation bias Bias
Seeking and weighing evidence that supports what you already believe. Leading questions to an AI are an easy way to get it without effort.
E
Error of commission
Acting on an incorrect automated recommendation.
Error of omission
Failing to notice a problem the automated system did not flag.
F
Fluency
How easily information is processed. In this course, a signal of familiarity, not of accuracy.
Framing effect Bias
Logically equivalent descriptions producing different judgments or choices. Applies to prompt wording as well as human decisions.
H
Heuristic
A mental shortcut that usually works and sometimes produces predictable errors.
I
Illusory truth effect Bias
Repeated statements feel more true because repetition makes them easier to process.
O
Opposite-framing test
Asking the same question in reverse (for example, risky? then safe?) and comparing the answers to see whether wording drives the result.
Overconfidence Bias
Confidence that exceeds accuracy. In AI use it is often driven by length, detail, and an unhedged tone.
P
Preprint
A research paper posted before peer review. Several sources in this course are preprints and are marked as such.
R
Representativeness heuristic Bias
Judging probability by how well something matches a prototype or stereotype, rather than by how common the outcome is.
S
Substitution Bias
Answering an easier question than the one asked, without noticing the swap.
Sycophancy Bias
A model's tendency to agree with the user's stated or implied beliefs, even when they are mistaken.
System 1
Fast, automatic, intuitive thinking. Runs continuously and produces impressions with little effort.
System 2
Slow, effortful, deliberate thinking. Supposed to check System 1, but often accepts its output.
System 3
Shaw and Nave's term for AI as a third cognitive system: external to the brain, fast, and able to shape how System 1 and System 2 operate.
W
WYSIATI Bias
"What you see is all there is." Judgments are built from the information at hand, with little weight given to what is missing.