Dispossession of the Educational Subject

Thinking with AI  ·  March 2026  ·  concept note

Dispossession of the educational subject names the process by which pupils and teachers gradually lose part of their intellectual autonomy and decision-making power to algorithmic systems. This is not a spectacular collapse, but a drift. Gestures that are first delegated only occasionally slowly become less exercised, and therefore less available: memorising, formulating, searching, explaining, ranking, judging. In the perspective developed by A Digital Ethic, the main danger is not only machine error. It is the quiet displacement of faculties that used to belong to the formation of the subject.

A soft and almost invisible loss

Today dispossession often takes the reassuring form of help. A student may think: “I do not need to learn this, I can ask AI.” A teacher may be tempted to delegate preparation, reformulation, evaluation or orientation to tools that promise speed, personalisation and efficiency. At first sight, nothing here looks like violence. And yet something shifts. What used to involve an inner appropriation is gradually externalised to systems that obey logics other than those of formation.

This dispossession is all the more troubling because it remains fluid. The more efficient the tool, the harder it becomes to perceive what its use takes away in the long run. Convenience says nothing about its cost. A faculty that is rarely mobilised weakens. A judgement that is constantly assisted loses strength. A thought that no longer has to cross hesitation, inquiry and effort deprives itself of part of its own formation.

The question is therefore not whether tools should be rejected, but from what point help ceases to be support and becomes a durable transfer of functions and criteria to the machine. Dispossession begins where the subject no longer uses the tool in order to think better, but gradually gets used to thinking less without it.

Four forms of dispossession

The first form is cognitive. Memory, calculation, formulation, information retrieval or first-draft writing are more and more easily delegated. These occasional delegations can be useful. Yet when they become systematic, they sustain a paradox: tools presented as augmenting our capacities may in fact atrophy the ones we no longer exercise. The more the machine retrieves, summarises or reformulates in our place, the less we maintain the skills required to do without it.

The second form concerns judgement. We tend to trust the answer that comes first, the ranking of a search engine, the synthesis of an AI, the score or signal produced by a platform. These systems, however, are crossed by design choices, data biases, commercial priorities or statistical consensuses that remain largely opaque. Dispossession is at work here in the transfer of critical evaluation to black boxes: we receive a result without always being able to trace the criteria that produced it, contest it or reinterpret it.

The third form is pedagogical. Some platforms promise personalised pathways, adaptive progressions and individualised recommendations. They may provide useful support. But they also tend to reduce learning to what is easily quantifiable, segmentable and automatable. Education, however, does not merely mean optimising a pathway. It means forming a relation to knowledge, time, effort, uncertainty and the common world. When contents, rhythms and even criteria of success are largely piloted by systems, teachers and students lose part of their deliberative power over what is worth learning and how it should be learned.

The fourth form concerns orientation and decision. The case of Parcoursup is an emblematic symptom. Trajectories may be distributed through complex mechanisms that remain difficult to read from the student's perspective. The subject does not only face an external decision. They experience an orientation partially mediated by systems they do not fully understand. Dispossession thus becomes institutional: it no longer touches only individual thinking, but the capacity to recognise oneself as the author of a path, a choice, a school destiny.

Why school is directly concerned

School is directly concerned because its task is not simply to provide ready-made answers, but to form subjects capable of seeking, interrogating and judging them. If learning is reduced to the immediate availability of a solution, access to information is confused with formation. Yet an educational subject is not formed by the mere accumulation of contents. It is formed in an active relation to knowledge: slowness, reformulation, confrontation with difficulty, explanation, revision and discernment.

Inside the cognitive multiverse, this requirement becomes even more decisive. Human, synthetic and hybrid contents circulate together, provenance blurs, and appearances of coherence multiply. Educating in such a world requires preserving spaces where one still learns to distinguish an answer from a judgement, a synthesis from an understanding, a suggestion from a decision. This is why digital metacognition becomes essential: it helps us perceive what tools do to our attention, our effort, our trust and our way of evaluating.

This dispossession is not equally distributed. Some will learn to use systems, understand their logics and keep a critical distance. Others will mainly get used to receiving answers, following recommendations and depending on permanent assistance. This uneven distribution opens onto what I call a two-speed cognitive justice. The issue is therefore not merely technical. It is social, educational and political: who learns to steer the tools, and who learns to live under their hold?

Reclaiming the subject

Responding to this dispossession does not mean banning tools or handing educational work over to them. It means rebuilding conditions in which the subject can remain the author of their thought. This requires preserving zones of friction: sometimes writing without assistance, explaining a line of reasoning aloud, justifying a source, comparing several answers, identifying what an AI reformulates without understanding, distinguishing what it helps us do from what it makes disappear. The aim is not technical purity. It is the preservation of judgement.

This also requires more lucid uses of artificial intelligence. An AI may help clarify an intuition, formulate an objection, explore a problem, open a path. That is precisely what dialogic exploration is about: stimulating thought without substituting itself for it. The right question is not: is AI present? The right question is: who keeps control over the criteria, the rhythm, the decision and the meaning?

Reclaiming the educational subject finally points toward a broader horizon: responsible digital autonomy. Not the mythical autonomy of an isolated individual, but the capacity to choose one's mediations, understand their effects, maintain an active relation to reality and avoid handing over without examination to opaque systems what falls under our responsibility. To educate in the age of AI is not only to teach how to use tools. It is to teach how not to be dispossessed by them.