AI: Why training a "well-crafted head" wasn't more critical

In an age of information overload, forming a "well-rounded head" is no longer about accumulating knowledge, but about learning to judge its depth, its validity, and its relevance to reality.

As artificial intelligence produces increasingly reliable texts, images and thoughts, an important educational question has arisen: What is the meaning of knowledge?In the era of information overload, building a "good head" no longer involves accumulating knowledge, but learning to judge its depth, its utility, and its significance in reality.

Today, students can produce work with impeccable form: structured, well-argued, sometimes brilliant.However, when we question them, anxiety arises.They struggle to explain what they really understood, to justify their choices, to relate what they produced to an experience or a concrete situation.Generative artificial intelligence (AI) is not always the direct cause of this situation, but it is a strong indicator of it.Because if generating information has never been so simple, understanding our operations has never been so demanding.

Know, know, understand: a distinction that has become central

In the age of artificial intelligence, the question of education is not considered in terms of knowledge acquisition.We should clarify what we mean by knowledge, knowledge and understanding and ask how this dimension is described in the educational process.

Two major epistemological traditions illuminate this distinction.Scientist Michael Polanyi has shown that all human knowledge has an immutable tacit element. That is, knowledge is rooted in the subject's experiences, actions, and commitments."We know more than we say," he says, emphasizing that understanding precedes clear formulation.This practical knowledge, often implicit, is built up through practice, trial, error, and encounter with reality.

Conversely, the philosopher Gaston Bachelard found that scientific knowledge does not simply arise from the elaboration of experience.This requires a break from primary evidence and the cost of opinion, reasoning, critical and abstract construction work."Science does not arise from ideas," he recalled, and it was necessary to train the mind to raise problems, not to accumulate answers.

The formation of a "well-formed head" therefore implies both the accumulation of abstract knowledge and the satisfaction of raw experience.It is learning to keep these two sides together: live experience and conceptualization, action and flexibility.

What AI can and cannot do

Artificial intelligence systems excel in areas where knowledge can be formalized: calculation, synthesis, multiplication, formatting.They occupy a growing share of accurate, consistent, identifiable knowledge.But they work in certain conditions: the correlation of statistics and the production of reliable information.

Artificial intelligence does not know or understand the world.He has no experience, no embodied relationship to reality, no access to the terms of the multiplicity of phenomena he describes.The information it generates is inherently probabilistic - it relies on probability calculations generated by statistical correlations rather than the understanding of causes, conditional - it depends on data, discourse contexts and technical standards, and reviewable - which means that it can be corrected, contradicted, or reformulated at any time without implying an internal evolution in understanding.

This difference is now at the heart of modern work on the educational use of AI, which shows that the automation of certain cognitive tasks, if not properly controlled, can weaken the use of critical judgment.

The more powerful the AI ​​works, the more likely it is to misrepresent the formal mix with transparency, ie.realism instead of careful conversation that opens the conversation.

Measuring complexity: a learnable skill

Faced with this situation, a major educational problem arises: the ability to measure the complexity of things.Distinguish what is on the informational surface from what a structured understanding implies.Assess the depth of a problem, system or situation.

However, this ability cannot be determined.It is gradually created through the experience of reality.It assumes an active work of conflict between what is theoretically assumed and what the test of concrete realization shows.It is in the gap between model and experience that the standards of judgment are refined, and it develops true macrointelligence in the sense that it clarifies formal knowledge and empirical knowledge.

Knowledge is effective only under conditions where it is tested, brought into tension with reality, and readjusted in the light of resistance and surprise.In contrast, raw experience is silent and difficult to communicate unless it is included in a reflective, conceptual framework.Training must therefore constitute the conditions for these difficult cycles between theory and practice, abstraction and implementation.

education and technological revolution

Understanding our devices doesn't just change.It defends the heart of higher mental functions: external memory, quick access to information, development of ideas.Earlier technologies enhanced already established human capabilities, and AI will redress its balance.

Accordingly, the educational challenge shifts.It is no longer a primary question of learning to produce or reproduce information, but to learn to assess its depth, coherence, validity conditions and effects in reality.This mutation is consistent with sociologist Edgar Morin's analyzes of complex thinking, which emphasize the need to train minds capable of relating, contextualizing, and dealing with uncertainty rather than reducing reality to simplistic answers.

Recent work in scientific research and educational science has shown that the use of AI can lead to more mental models, reducing the use of memory and long-term memorization, while use and application can strengthen learning.

Training engineers - and citizens - who can judge

Forming a well-formed head in the age of AI means not confusing cognitive representation and intellectual abandon.It is about training subjects capable of using powerful systems without subjugating them, capable of supporting the need for meaning where the machine only produces form.

At IONIS, the development of the IONIS Institute of Technology (I2T) on our campuses stems from a strong belief: if our engineering students master AI technology, they must also learn to test their limits by facing the real world.The laboratory, the workshop and the experiments then become central places for the formation of judgement.

Educating good engineers—and, in a broader sense, enlightened citizens—consists of cultivating a critical, measured, and evolving spirit that is nourished by concrete experiences of how to work and how to disrupt.Therefore, in the age of artificial intelligence, the main question is not only what we expect from machines, but also what we expect from people: their ability to understand, create and make decisions in an uncertain and technologically enhanced environment.