Inquiry, in Dewey's precise usage, is not ordinary thinking but the disciplined process by which the organism resolves genuine doubt. It begins with a felt difficulty — a situation whose existing meanings no longer serve. It proceeds through the formulation of a problem, the generation of hypotheses, the reasoning-through of implications, the testing of hypotheses by action, and the reconstruction of the situation and the inquirer's understanding in light of the result. Each phase is essential. Each takes time. The phases cannot be compressed without altering the character of what is being done. Inquiry is intelligence in its operational form, and the question the AI age forces is whether the conditions of AI-augmented work preserve the phases of inquiry or eliminate them in the name of efficiency.
Dewey laid out the structure of inquiry with the care of an engineer describing load-bearing members. The felt difficulty occasions the whole process; without genuine doubt, nothing motivates thought. The definition of the problem transforms the undifferentiated trouble into something that can be addressed. The generation of hypotheses brings the inquirer's own experience to bear on possible resolutions. The reasoning-through of implications tests each hypothesis mentally before committing to action. The testing through action confronts the hypothesis with the resistance of the situation. And the reconstruction integrates what was learned into the organism's ongoing understanding, changing the inquirer as much as the situation.
Each phase is threatened in different ways by AI-augmented work. The felt difficulty is diminished when plausible answers are available immediately — why doubt when the machine is confident? The definition of the problem is short-circuited when the tool proposes solutions before the problem has been fully articulated. The generation of hypotheses atrophies when AI provides hypotheses before the inquirer has generated her own. The reasoning-through is compressed when the interval between description and result collapses to seconds. The testing through action becomes evaluation of machine output rather than direct encounter with the domain. The reconstruction requires reflection, which requires time, which the temporal logic of AI-augmented work does not provide by default.
What survives is the appearance of inquiry. Problems are stated. Solutions are produced. Outputs are evaluated. But the educational substance — the modification of the inquirer's understanding through the full cycle — may have been bypassed. The Deleuze episode Segal recounts in The Orange Pill is an instance of inquiry reasserting itself: he received a plausible passage, noticed something that nagged, sat with the doubt, investigated, and reconstructed his understanding of what the tool could and could not be trusted to produce. Inquiry did not happen automatically. It happened because he slowed down.
Dewey's most complete account of inquiry appears in Logic: The Theory of Inquiry (1938), the culminating synthesis of a lifetime's work on how thought actually operates. The earlier How We Think (1910, revised 1933) developed the five-phase structure for educational audiences; Logic generalized it into a theory of all genuine knowing.
Five-phase structure. Felt difficulty, problem definition, hypothesis generation, reasoning-through, and testing — each essential, each time-consuming, each vulnerable to compression.
Genuine doubt is the engine. Without a felt difficulty, the machinery of inquiry has nothing to run on; manufactured problems produce the appearance of inquiry without its substance.
Reconstruction is the point. The outcome of inquiry is not a conclusion stored in memory but a changed inquirer — modified dispositions, widened perceptions, new capacities.
Inquiry has temporal thickness. It cannot be executed at machine speed without altering what it is; the interval between phases is where the cognitive work happens.