## 1. A change that entered through the back door

Generative artificial intelligence did not enter schools through policy, curricular reform, or a clear strategic decision. It entered through students’ everyday practices: homework, test preparation, and last-minute explanations before an assessment. It entered quickly, diffusely, and irreversibly.

Meanwhile, schools largely continue to operate as if nothing structural had changed. This growing mismatch between real practices and pedagogical models is now the central problem of education.


## 2. When answers no longer prove thinking

Today it is possible to produce correct answers, fluent texts, and acceptable solutions without going through formative cognitive processes. When this happens, what is at stake is not only assessment. It is the very meaning of learning.

For decades, secondary education rested on an implicit assumption: to answer, write, or solve, students had to think. That assumption is no longer valid. Continuing to assess as if it still were is not rigor. It is a simulation of rigor.

The core problem is not individual misconduct, nor technology itself. It is the loss of correspondence between *output* and learning. When schools certify products without being able to justify the processes behind them, pedagogical authority erodes.


## 3. Why secondary education is the critical point

The impact of artificial intelligence is particularly severe in secondary education. This is the stage where habits of thought are consolidated, relationships with effort and error are formed, intellectual authorship develops, and academic and vocational decisions with long-term consequences are prepared.

Integrating AI without explicit pedagogical framing tends to encourage early delegation of thinking, cognitive superficiality, invisible dependency, and the amplification of inequalities between students with different cultural backgrounds. The most serious effects do not appear immediately in grades. They emerge later, when there is no longer sufficient internal structure to learn something new without external support.

Beyond the family, secondary education may be the last space where society can still intervene systematically in the formation of thinking. Failing to act here is an error with irreversible consequences.


## 4. Teaching is not education

It must be said clearly: this is not a text against artificial intelligence. AI does not destroy education. What it does is make existing fragilities unsustainable.

Artificial intelligence has a profound impact on **teaching**: it supports explanation, practice, production, and clarification of content. Its impact on **education**—the formation of values, character, empathy, responsibility, and judgment—is limited and indirect.

The risk arises when pedagogical efficiency is confused with human formation. A school can become faster and more productive while becoming poorer in its educational mission. The more AI occupies the space of teaching, the more consciously schools must protect the space of education.


## 5. The principle that organizes everything: what can and cannot be delegated

At the center of this reflection is a simple and demanding founding principle:

> **In secondary education, what does not constitute the formation of thinking may be delegated.**  
> **What builds intellectual autonomy, identity, and responsibility must be protected.**

Thinking before answering, explaining in one’s own words, taking a reasoned position, defending what one claims, and assuming responsibility for consequences are not pedagogical details. They are the irreducible core of learning. If these operations are delegated, the result may be acceptable. But the student is not formed.


## 6. An agenda for deciding, not postponing

What is proposed here is not technical or instrumental. It is a pedagogical and institutional agenda for the period 2026–2030, organized around six strategic axes:

- thinking before response  
- literacy as judgment, not training  
- authorship as intellectual appropriation  
- process-centered assessment  
- ethics as operational practice  
- school as a space of deceleration  

Restoring rigor and educational justice in an AI-rich context does not require more technological surveillance. It requires greater pedagogical responsibility. Assessment shifts from checking final products to making thinking visible. School governance shifts from reacting to cases to deciding criteria.


## 7. Two documents to deepen and act

This work is available in two complementary formats:

- **Executive Summary**  
  Addressed to educational decision-makers and school leadership, presenting the problem, principles, and key criteria to support conscious decisions.

- **Full Document**  
  An in-depth essay developing the foundations, risks, and implementation proposals for secondary education.

Neither aims to close the debate. The goal is to make it governable.


## 8. An invitation to responsibility

In a time of acceleration, delegation, and externalization of thinking, pedagogical governance is the true educational act. Educating today means deciding clearly what schools protect, what they delegate, and what they certify as legitimate learning.

This text is an alert, a reflection, and an invitation. Reading the documents is the next step. Deciding and acting is what follows.