Phronesis (Flyvbjerg's Reading)

In the AI Story

The distinction from episteme and techne is categorical, not gradient. Episteme is context-independent; the law of gravity does not vary by jurisdiction. Techne is rule-governed; a master carpenter can teach joinery through instruction and practice. Phronesis is both context-dependent and not reducible to rules — it requires the reading of a particular situation that no general principle can supply. The phronimos, the person of practical wisdom, does not merely know what is true or what can be made; the phronimos knows what should be done, which is a question neither episteme nor techne can answer.

AI systems, as currently designed, are episteme-and-techne machines of extraordinary power. They process universal knowledge — the statistical regularities extracted from vast training corpora — with speed and accuracy that exceed human capacity. They produce technical artifacts — code, text, images, analyses — with a facility that makes the most skilled human practitioner seem glacial. What they cannot do is exercise phronesis. They cannot weigh competing goods when the competition is between incommensurable values; they cannot bear responsibility for outcomes that cannot be predicted; they cannot make the kind of situated, value-laden judgment that Aristotle identified as the architectonic virtue.

Aristotle's claim — formulated in the fourth century BCE — was that phronesis is architectonic: the virtue that governs the exercise of all other virtues, that determines when and how and for what purposes the other forms of knowledge should be deployed. Without phronesis, the other virtues are blind — powerful but directionless, capable but ungoverned. This ancient claim turns out to be the most precise description available of the AI transition's central challenge. The machines have episteme and techne in abundance. What governs their deployment is the phronesis of the humans who direct them.

The practical implication is that a civilization organized around episteme and techne — as Western civilization has been since the Scientific Revolution — has systematically neglected the virtue most needed in the age of AI. Educational pipelines, hiring processes, performance reviews, career ladders: all organized around the epistemic-technical virtues, all blind to phronesis. The AI transition has exposed this systemic blindness with brutal clarity. When the tools automate the techne, the practitioners who thrive are the ones who happened to develop phronesis along the way. Their development was not planned. Their capacity was not trained. Their value was not recognized until the machine made it impossible to ignore.

Origin

Flyvbjerg's rehabilitation of phronesis began with his 1998 Rationality and Power, the Aalborg case study that established phronetic social science as a methodological alternative. The theoretical framework received its fullest statement in Making Social Science Matter (2001). The 2025 paper 'AI as Artificial Ignorance' extended the framework to artificial intelligence, identifying phronesis as the capacity AI systems structurally cannot possess.

Key Ideas

Architectonic virtue. Phronesis governs the exercise of all other virtues, determining when and how episteme and techne should be deployed.

Context-dependent. Unlike episteme, which holds across contexts, phronesis requires the particular reading of a particular situation with its particular stakes.

Not rule-governed. Unlike techne, which can be taught through instruction, phronesis develops only through sustained engagement with situations that resist rules.

Structurally absent in AI. Current systems possess episteme and techne in abundance and cannot exercise phronesis because they lack the embodied stakes and situated judgment the virtue requires.

Institutionally neglected. Modern institutions are organized around the epistemic-technical virtues and have systematically failed to cultivate, recognize, or reward phronesis.

Debates & Critiques

Skeptics have argued that phronesis is a mystification — that what it names is simply expertise that has not yet been codified, and that sufficient analysis would reduce it to techne. Flyvbjerg's response, developed through decades of empirical work, is that phronesis persists precisely at the point where codification fails — in the context-dependent judgment that no rule system can capture. The AI transition, by demonstrating what can and cannot be automated, provides empirical support for the non-reductive reading.