AI in our schools: let’s teach our children to grow well with it

The question is no longer whether our children will use artificial intelligence. They already use it in their search engines, their phones, their apps, and their creative tools. It is gradually shaping the way they look for information, write, solve a problem, and understand a subject. Within a few years, it will probably seem as natural to them as the web, the smartphone, or the calculator became for previous generations.

We therefore have little to gain from imagining a school where artificial intelligence would be kept at the door. It will accompany young people through their studies and then into their professional lives. It will also keep becoming more accessible, more capable, and able to take part in a growing number of intellectual activities.

The real educational challenge is far more interesting: teaching young people to use an extraordinarily powerful technology while preserving the intellectual effort that allows them to develop their own knowledge, creativity, autonomy, and judgment.

Schools have already been through several technological transformations. Artificial intelligence, however, has one important particularity: it can carry out part of the intellectual work that is precisely the learning process. It can write, summarize, explain, calculate, argue, and solve certain problems. The question therefore becomes how to use this power to make students learn more rather than simply produce more answers.

A genuine artificial intelligence literacy is beginning to emerge in the world of education. UNESCO, in particular, has developed a competency framework for students that goes well beyond the ability to write good prompts. Understanding AI, using it responsibly, developing critical judgment, reflecting on ethical issues, and possibly learning to create with it are all part of a much broader preparation.

This distinction is essential because knowing how to use an artificial intelligence does not automatically mean knowing how to think. A student can ask an AI to summarize a novel they have not read, to solve a math problem they do not understand, or to write an excellent text on a subject they barely grasp. The result produced can be impressive while the learning remains almost non-existent.

The OECD draws attention to precisely this difference between performance and learning. General-purpose generative AI tools can improve the quality of the output students produce without that improvement necessarily corresponding to a lasting acquisition of knowledge. When the tool takes on too large a share of the cognitive effort, the student risks taking much less part in the reasoning that would have allowed them to progress.

This observation should guide the way we integrate AI into education. The objective should not be to measure how much work artificial intelligence can do in the student’s place. We should instead seek to determine how it can make their brain work harder and help them develop capabilities they will retain afterward.

Used in this way, artificial intelligence becomes a tremendous intellectual partner. A student can ask it to explain a difficult concept another way. They can present their reasoning and ask it to identify the weaknesses. They can compare two explanations, explore hypotheses, simulate a debate, or obtain additional exercises tailored to their difficulties.

A young person learning a language can have a conversation partner available almost at any time. One struggling with mathematics can ask for a new explanation after school. A particularly advanced student can explore a subject they are passionate about far more deeply, while another can ask for a concept to be explained in several ways until the wording that suits them finally makes it click.

The potential to democratize educational support is considerable. Forms of tutoring that previously required an adult’s availability can become far more easily accessible. This capability can be particularly valuable when it complements teachers’ work and allows students to continue learning between class periods.

It should, however, come with a fundamental intellectual rule: learning to ask for help without giving up on thinking. AI can explain while the student retains responsibility for understanding. It can suggest avenues the student must evaluate, critique reasoning the student must learn to defend, or propose an answer whose soundness the student must gradually become able to assess.

That last capability will probably become one of the most important. The more capable and convincing artificial intelligence becomes, the more we must preserve enough human skill to judge what it produces.

A perfectly worded answer can contain an error. A convincing explanation can overlook an essential nuance. A source can be misinterpreted, and perfectly logical reasoning can rest on an incorrect premise. The fluency of an answer never guarantees its truth.

The ability to detect these problems rests precisely on what schools have always sought to develop: knowledge, curiosity, reasoning, the ability to look for evidence, and critical thinking. The OECD stresses in particular the importance for students of continuing to develop their critical thinking independently of AI, of formulating and testing hypotheses, of navigating uncertainty, and of evaluating the quality of evidence.

This reality produces an interesting paradox. The more machines we have that can produce answers, the more valuable the human ability to recognize a good answer becomes.

We will therefore have to keep reading, writing, counting, memorizing certain knowledge, solving problems without assistance, arguing, doubting, and searching. These forms of learning do not become useless because a machine can now carry out some of these tasks. They are the foundations that will allow humans to work intelligently with it.

The calculator offers an imperfect but instructive analogy. Its existence did not eliminate the need to understand numbers. We still teach arithmetic because an individual must have enough understanding to know which operation to perform, interpret the result, and detect an obviously inconsistent answer. With artificial intelligence, this principle extends to a far broader share of intellectual activity.

School can therefore become one of the best places to learn to use AI properly. Since young people will encounter these systems anyway, the educational environment offers exactly the opportunity to teach them to experiment with support rather than letting them develop their usage habits on their own.

This education can include how to verify an answer, recognize an uncertain claim, protect personal information, look for sources, distinguish credible information from merely plausible wording, and properly disclose an AI’s contribution to a piece of work. It can also teach students to recognize the situations in which using AI is appropriate and those in which the educational objective is precisely to complete the exercise without it.

Learning to question AI could even become an exercise in critical thinking. Asking why an answer is being proposed, what evidence supports it, what the counterargument would be, which assumptions might be wrong, or how to approach a problem without immediately obtaining its solution forces students to think about the very structure of reasoning.

Artificial intelligence can then become a training ground for judgment rather than a way of avoiding it.

This transformation also reinforces the importance of teachers. When information becomes abundant and a machine can generate an explanation in seconds, the teacher brings a different dimension: they know the student, observe their progress, detect their difficulties, spark their curiosity, encourage them to persevere, and help them develop their judgment.

They also have the ability to design the learning experience. The same artificial intelligence can be used to bypass an exercise or to make it far richer. The difference often lies in how the activity is built and in what the teacher asks the student to do with the tool.

This evolution obviously requires that teachers themselves be supported. It would be difficult to ask them to prepare young people to work with artificial intelligence without giving them the time, knowledge, tools, and rules needed to understand this technology and experiment with it.

Developing genuine AI literacy must therefore concern both students and educators. Training initiatives can help create a common language that allows institutions to move gradually beyond first reactions focused solely on bans, detecting generated work, or fear of cheating.

These concerns remain legitimate, but they represent only part of the problem. The far broader challenge is to determine which human skills we want to develop in a world where part of the intellectual work can be entrusted to machines.

This question becomes even more important when we look beyond school. The children starting their schooling today will enter a labor market where collaborating with artificial intelligence will probably be as commonplace as working with computers is today.

Some tasks will change profoundly and others may disappear. New activities will emerge, while humans will work with assistants, copilots, and agents capable of carrying out part of the intellectual work. Organizations will have to learn to distribute responsibilities between people and systems according to their respective capabilities.

Preparing young people for this world therefore requires more than technical training in how to use AI. They will have to be able to understand a problem, choose the appropriate tools, use the available machines intelligently, check their work, make a decision, and take responsibility for that decision.

This capability will probably prove far more durable than mastery of a particular model or interface. The tools a child uses today may have changed completely by the time they enter the labor market. The intellectual habits they will have developed—learning, questioning, verifying, reasoning, creating, and exercising judgment—will keep their value.

We can thus envisage a generation whose capabilities will be considerably enhanced by artificial intelligence without its own intelligence being put on standby.

This perspective also matches a broader vision of transformation through AI. At Quantum Beyond, we believe the purpose of this technology is to augment human and organizational capabilities. This philosophy applies to companies learning to integrate agents into their teams, but it begins much earlier, with the people who will have to live and work in this environment.

A society that wishes to take full advantage of artificial intelligence therefore has every interest in simultaneously developing a population capable of understanding it. AI literacy is gradually becoming a civic skill, in the same way as the ability to understand digital information, protect one’s identity, recognize a credible source, and exercise judgment in an environment where technology is continuously involved.

Artificial intelligence arrived quickly, and education systems did not have several decades to prepare for its integration. It is already in students’ hands and will keep becoming more accessible, more capable, and more present in their daily lives.

This situation offers us a remarkable opportunity. For the first time, it becomes conceivable to give almost every young person access to a tool able to explain, translate, question, propose exercises, stimulate creativity, and support part of their learning practically on demand.

Our responsibility is to ensure that this power develops their capabilities rather than excusing them from acquiring them. That calls for genuine AI literacy, supported teachers, suitable teaching approaches, and a much finer understanding of the place we want to give these technologies in human development.

Success should therefore not be measured solely by students’ ability to use artificial intelligence effectively. It should also be visible in their ability to think without it when necessary, to recognize its limits, to question its answers, to understand subjects well enough to exercise their judgment, and to choose intelligently when using it brings real value.

This vision aligns deeply with that of Quantum Beyond. We believe in the extraordinary potential of artificial intelligence and in its adoption. Its greatest success, however, will never be measured solely by the number of tasks it can perform. It will also be measured by what humans become able to understand, create, and accomplish thanks to it.

For our children, the stakes are considerable. They will probably have intellectual tools of a power that previous generations could scarcely imagine. We now have the opportunity to make sure that, as they grow up with these technologies, they simultaneously develop their knowledge, creativity, curiosity, autonomy, and judgment.

The future we must prepare for them is therefore a world in which they will know how to live and work with artificial intelligence while retaining the capabilities that will let them decide when to trust it, when to question it, and when to think for themselves. If we succeed in this transition, AI can become much more than a shortcut to an answer: it can help shape a generation even better equipped to learn, think, create, and act.