CONCEPT
Anomaly Detection (Klein)
The expert capacity to register meaningful deviations from expected patterns — the hallmark of genuine expertise and the function most endangered by AI-mediated work.
Anomaly detection is Klein's term for the expert's ability to register that something in the current situation does not fit the pattern of normal she has built through extensive experience. It is the function that let NICU nurse Darlene pause at an infant whose monitors showed normal readings and detect early sepsis hours before equipment would have caught it. The capacity depends on a pattern library rich enough to generate precise expectations — because the precision of the expected pattern determines how subtle a deviation can be detected. Anomaly detection is also what Klein identifies as the most important human contribution in contexts where AI produces statistically plausible but occasionally wrong outputs. The AI generates output that conforms to training data; the human expert detects when conforming is not the same as being correct.
In The You On AI Field Guide
Darlene's detection of sepsis illustrates the architecture of anomaly detection. The infant's color was 'not right' — not the pink flush of a healthy infant but something she