Published during an earlier fever cycle of artificial intelligence, the Dreyfus brothers argue that novices may depend on explicit instructions, but genuine expertise develops in the opposite direction.

That is, towards situated judgement, pattern recognition and an embodied sense of what matters before conscious calculation has caught up. Their influential five-stage model traces this movement from novice through advanced beginner, competence and proficiency to expertise. 

The distinction at the book’s centre is between knowing that and knowing how. A computer may process formal descriptions with enormous speed, but the authors contend that skilled people do not usually experience the world as a neutral field of data. The experienced nurse, driver, chess player or craftsperson perceives a meaningful situation already arranged by relevance. Expertise is not mystical intuition floating above reason. It is accumulated contact with reality compressed into perception.

Some of its predictions about computer capability belong unmistakably to 1986, but the underlying argument has become more interesting rather than less. Contemporary machine learning can imitate patterns that earlier symbolic systems could not, yet the book’s deeper questions remain open: whether statistical performance amounts to understanding, whether knowledge can be detached from participation, and what is lost when institutions mistake what can be measured for what can be known. An excellent historical counterweight to both AI evangelism and the equally lazy claim that computers are merely tools. 

First edition, Free Press, 1986. Hardcover with dust jacket. Jacket rubbed with light scratching, edge wear and small chips/tears at the lower corners; boards and pages appear clean and sound.