What Should Students Still Learn?
Curriculum in the age of intelligent systems
The question behind the question
If artificial intelligence can answer, what is the value of knowing? It is a deceptively simple question, but it cuts through the entire structure of education. For centuries, universities have been organised around the transmission of knowledge. Students learned what was known, reproduced it in different forms and were assessed on their ability to recall, apply and extend it.
That model rested on a stable assumption: knowledge was scarce, access required effort and understanding required time. That assumption no longer holds.
When answers come first
Students now operate in an environment where answers are immediate. Explanations can be generated, concepts summarised and problems solved step by step—often before the student has fully understood the question itself.
This creates a subtle but fundamental reversal in the learning process.
Where education once moved from question to exploration to answer, it now increasingly risks moving from answer to interpretation to partial understanding. Students are no longer working towards answers; they are working backwards from them.
The sequence has changed. And with it, the role of the student.
From knowing to navigating
If knowledge is always available, education can no longer be organised around possession. It shifts towards navigation.
The central question is no longer what students know, but how they approach what they do not yet understand. This requires a different kind of intellectual posture—one that is comfortable with uncertainty and capable of working through it.
Students must learn to frame problems before solving them, to recognise what is missing in an answer and to question the assumptions embedded in generated output. In a world of abundant answers, the quality of the question becomes decisive.
AI takes tasks, not responsibility
This is the core distinction. Artificial intelligence can generate, structure and optimise. It is an engine of output. But it does not define the problem, choose the objective or carry the consequences. It does not take responsibility.
Human beings do.
As AI systems become more capable, this distinction becomes more important, not less. The role of education is therefore not diminished, but clarified. It is no longer centred on producing answers, but on forming the capacity to decide which answers matter—and what to do with them.
Judgment as the core capability
This leads to a redefinition of the curriculum. Not as a list of subjects, but as a structure of capabilities, all of which ultimately contribute to one overarching capacity: judgment.
Judgment is the ability to connect knowledge to action. It determines what is relevant, what is reliable and what is acceptable. It cannot be automated, because it is inherently tied to responsibility.
Within this broader frame, several capabilities become central.
Learning to define the problem
The first is problem definition. Not all problems are given clearly and not all questions are well-formed.
Students must learn to clarify ambiguity, identify relevant dimensions and distinguish between symptoms and causes. This is not a preliminary step, but the foundation of meaningful work.
In an environment where answers are easily generated, poorly defined questions become a structural risk.
Critical thinking under conditions of abundance
Critical thinking becomes more concrete in an AI-mediated context. It is no longer an abstract academic ideal, but a practical necessity.
Students must be able to evaluate outputs, detect inconsistencies and recognise when coherence is superficial rather than substantive. AI can produce text that reads convincingly, but plausibility is not the same as correctness.
The task is not to reject AI-generated output, but to interrogate it.
Contextual literacy
As AI systems synthesise information without exposing their full reasoning, understanding context becomes essential. It is no longer sufficient to know what is said; students must understand why it is said, how it is constructed and what its limitations are.
This form of literacy extends beyond sources. It includes awareness of how knowledge is produced, how models are trained and where blind spots may exist.
Trust, in this environment, is not given. It must be actively constructed.
AI literacy as understanding, not usage
AI literacy, therefore, is not about mastering tools, but about understanding their nature.
Students need to grasp that AI systems generate probabilistic outputs rather than definitive truths. They must understand the role of training data, the presence of bias and the limits of machine reasoning.
Without this awareness, AI risks becoming an invisible authority. With it, AI remains a tool that can be questioned, interpreted and, when necessary, resisted.
Epistemic humility
In a world where answers are always available, knowing what one does not know becomes a critical skill.
Students must develop what might be called epistemic humility: the ability to recognise uncertainty, to question apparent certainty and to remain open to revision. This is not a weakness, but a form of intellectual discipline.
AI systems rarely express doubt. Education must teach students to do so.
From analysis to synthesis
Artificial intelligence is highly effective at analysing and decomposing information. It can break problems down, identify patterns and generate structured responses.
The human contribution shifts towards synthesis.
Students must learn to connect domains, to bring together insights from different fields and to construct meaning across contexts. This is where originality increasingly resides—not in producing isolated answers, but in combining them in ways that reveal something new.
Embodied knowledge
Not all knowledge is abstract or textual. In fields such as healthcare, education and social work, knowledge is also embodied. It is expressed in interaction, in timing, in empathy and in situational awareness. These forms of knowing cannot be fully captured in language, nor can they be generated by a system.
A curriculum that neglects this dimension risks becoming detached from practice.
The future of education must therefore include not only cognitive skills, but also the capacity to act in real-world contexts where judgment is exercised under conditions of uncertainty.
The illusion of efficiency
There is a strong temptation to redesign education around efficiency. Faster learning, more output and smoother processes appear attractive. But learning has never been efficient.
Understanding often emerges slowly, through repetition, confusion and eventual clarity. The moment when a difficult idea finally “lands” cannot be accelerated indefinitely. It depends on engagement, not optimisation.
Artificial intelligence removes much of the friction that traditionally accompanied learning. Yet it is precisely this friction that makes learning possible.
A curriculum that optimises for speed risks producing answers without depth and competence without grounding.
The student as operator—or thinker
This leads to a fundamental choice. Do we educate students to operate systems or to think independently of them?
The first path produces efficiency. The second produces resilience.
Students who rely on systems for reasoning may perform well under normal conditions. But when systems fail or produce flawed outputs, they need the capacity to recognise and correct those failures.
That capacity does not emerge automatically. It must be taught.
Designing the curriculum differently
If education shifts from knowledge transfer to capability formation, curriculum design must follow.
This implies less emphasis on reproduction, standardised answers and isolated assignments and more emphasis on open-ended problems, interdisciplinary thinking and iterative work.
Students should not only produce answers, but also explain how they arrived at them, what uncertainties remain and how they would proceed if those uncertainties could not be resolved.
Assessment, in this sense, becomes an extension of learning rather than a separate mechanism.
Europe’s position
Europe occupies a distinctive position in this transition. It does not dominate the infrastructure of AI, nor does it set the pace of technological development.
But it has a long-standing tradition of critical inquiry, institutional accountability and public responsibility.
If these principles are embedded into curriculum redesign, Europe can shape a different model of education—one that does not simply adapt to intelligent systems, but defines how they are integrated into human learning.
The deeper shift
The question is not whether students should still learn facts. They should. But facts are no longer the endpoint. They are the starting point.
The deeper shift moves from knowing, to understanding, to judging. From information, to interpretation, to responsibility.
Education is no longer about acquiring answers. It is about developing the capacity to live with them—and to question them when necessary.
Final line
If AI can answer almost anything, the value of education lies in knowing when not to accept the answer.
This article is part of the series The University After AI, published in the Culture & Education section of Altair Media.
Photo by Hadija on Unsplash
When answers are only a kilometre away, the value of education lies in knowing which direction to take.
