Publish at March 11 2026 Updated March 11 2026
In a simulation room, a paramedical student analyzes a complex clinical case. The data are numerous: biological constants, antecedents, symptoms. Even before formulating his initial hypotheses, he queries a decision support system. In a matter of seconds, a summary appears, ranking possible diagnoses, suggesting courses of action and prioritizing risks. The time saved is obvious. The relevance is often remarkable. The tool doesn't replace the decision, it guides it. And this is precisely where the question begins...
Artificial intelligence is now being integrated into clinical practices and healthcare training systems:
The rise of generative models has accelerated this presence. In educational environments, these tools promise individualized support, greater exposure to complex cases and rapid, structured feedback. At first glance, AI appears to be an amplifier of skills.
But training in the age of decision support systems is not just about adding a tool to the pedagogical toolbox. It means accepting that the very conditions of clinical reasoning need to be reconfigured. Professional reasoning in healthcare is not simply a matter of aggregating data. It presupposes a capacity for hypothetization, a tolerance for uncertainty, a confrontation with doubt, and an assumption of responsibility for the decision taken. It involves a professional subjectivity built into the effort of interpretation.
However, when an algorithmic suggestion precedes or frames the analysis, the cognitive order is subtly modified. The hypothesis no longer emerges solely from the clinical examination; it can be triggered, confirmed or restricted by the tool's suggestion. The risk is not that the algorithm will be wrong: error is inherent to all clinical practice. The risk is more discreet: that the effort to think will be reduced, that the student will become accustomed to validating rather than constructing, that assistance will become the implicit matrix of decision-making.
This is the tension that runs through healthcare training today. Far from being neutral, digital devices redistribute forms of power and regimes of responsibility. As the analyses of technology by Gilbert Simondon and Ivan Illich have shown, a tool is never a simple instrumental extension: it configures an environment. It shapes gestures, expectations, rhythms, and even modes of professional subjectivation.
In a regulated context, where decisions involve the civil, criminal and ethical responsibility of the caregiver, this reconfiguration cannot be ignored.
The issue, then, is not to pit technological innovation against the ethics of care. It's about understanding how to integrate AI without silently shifting the center of gravity of professional judgment. Training with AI does not mean training under AI. In the first case, the tool supports explicit reflection; in the second, it becomes the implicit reference around which decisions are organized.
The hypothesis defended here is as follows: AI does not replace clinical decision-making, but it does transform the conditions in which it is developed. In training, the major danger is not algorithmic error, but the progressive delegation of reasoning. The challenge then is to design systems capable of preserving the effort of hypothetizing, maintaining the capacity for doubt, and maintaining the full responsibility of the future professional.
Can we learn to decide if we no longer learn to tolerate uncertainty? Does AI enhance the quality of clinical reasoning, or just its apparent speed? And how can we train autonomous caregivers in an environment where algorithmic suggestion is becoming constant?
Clinical decision support systems are by no means an absolute novelty. As early as the first computerized medical databases, the ambition was already to support the practitioner in interpreting complex data. What has changed today is the scope and plasticity of these systems. Statistical inference engines have given way to architectures capable of synthesizing massive corpora, generating differential hypotheses, anticipating risks or reformulating reasoning in a pedagogical language.
This evolution is particularly visible in training environments. Simulation platforms integrate conversational agents capable of playing the role of standardized patients. Generative tools propose case analyses, reformulate reports and simulate interprofessional exchanges. Students can question a system on the plausibility of a diagnosis, request a summary of recommendations or explore alternative scenarios in a matter of seconds.
These pedagogical uses are not to be confused with the actual practice of care. In training, the tool can be mobilized as a learning support, an exploration resource or a discussion catalyst. Yet the boundary is porous. The student trained with assistance will be the professional who practices with assistance. The cognitive habits acquired in the pedagogical setting do not disappear when the student takes up the job. In other words, the way AI is integrated into training already prefigures a certain configuration of professional judgment.
The promise is seductive: exposure to an increased diversity of clinical cases, rapid feedback, personalized pathways, instant access to up-to-date recommendations. AI appears to amplify human analytical capacity. It broadens the information horizon and reduces the memory load. But all amplification also transforms the structure of the effort required.
The analytical power of these systems lies not only in their technical performance, but also in the way they are perceived. Algorithms benefit from a particular symbolic capital: they embody massive data, aggregated statistics and supposed objectivity. Where human reasoning can appear fragile, situated, influenced by cognitive biases, the machine appears distant and rigorous.
This perception gives rise to a phenomenon ofalgorithmic authority. The suggestion proposed by the system is not imposed by explicit constraint, but by implicit credibility. Faced with a hierarchical list of diagnoses or a well-argued course of action, the student may find it difficult to challenge the proposal. Not because they are incapable of doing so, but because the tool is presented as the synthesis of a superior rationality.
The effect is all the more powerful in that digital environments tend to naturalize their recommendations. The interface is fluid, the response rapid, the formulation assured. Nothing signals the uncertainty inherent in the underlying probabilistic models. The suggestion is self-evident. A discrete shift in the decision-making center of gravity: the hypothesis no longer arises exclusively from clinical examination, but emerges in dialogue with a technical device.
From this perspective, AI doesn't just assist; it frames. It directs attention towards certain elements, makes others invisible, and prioritizes possibilities. Like all devices in the Foucauldian sense, it organizes a field of visibility and a regime of truth. The human decision remains formally intact, but is elaborated in an already structured space.
In the healthcare professions, decision-making involves a high degree of individual responsibility, enshrined in a precise legal and ethical framework. However, the massive introduction of algorithmic tools is reconfiguring the effective distribution of decision-making power. Knowledge no longer resides solely in the memory and experience of the professional; it is outsourced to databases, predictive models and proprietary systems designed by industrial players.
This outsourcing raises a number of questions. Who really understands the tool's internal logic? Who masters the weighting criteria, training data sets and validity limits? Students, trainers and clinicians rarely have complete transparency. The tool is used in a regime of partial trust, often asymmetrical.
In training systems, this asymmetry is further accentuated. Students discover systems whose results they perceive without knowing their architecture. They learn to interact with a machine whose logic remains opaque. AI thus becomes a shared technology: it circulates between designers, institutions, professionals and learners, subtly redistributing the margins of autonomy.
This does not mean outright dispossession. The tool can reinforce vigilance, suggest diagnoses and prevent errors. But it transforms the way decisions are made. Authority does not disappear; it shifts. The professional remains legally responsible, while the tool silently influences the way judgments are made.
Thus, clinical AI presents itself as an undeniable analytical power, capable of enriching learning and supporting decision-making. It broadens access to knowledge, speeds up synthesis and opens up unprecedented training prospects.
But with this power comes cognitive seduction. Algorithmic authority, fluid interfaces and rapid responses are changing the very conditions of intellectual effort. What is facilitated can also be unburdened. What is assisted can become anticipated.
So the question is no longer simply one of the tool's performance, but of the habit it establishes. At what point does assistance cease to be a support and become a crutch? At what point does consultation become dependency? It is precisely this discreet but structuring shift that we must now examine: the gradual cognitive delegation of clinical reasoning.
If AI amplifies analytical capacity, it also transforms the nature of the cognitive effort required. All assistance modifies the balance between what is produced by the subject and what is provided by the device. In training, this balance is decisive: learning to exercise a healthcare profession is not simply a matter of accessing relevant answers, but of constructing a judgment capable of supporting itself. When assistance becomes constant, the risk is not technical error, but the gradual erosion of intellectual autonomy.
Clinical reasoning is traditionally based on a hypothetico-deductive model: faced with a given situation, the professional formulates hypotheses, gathers additional data, eliminates possibilities, refines his diagnosis and justifies his decision. This process is neither linear nor mechanical. It is permeated by uncertainty, interpretative tensions and successive adjustments.
Health training is based precisely on learning about this dynamic. Students learn to tolerate doubt, to accept that they don't know immediately, and to postpone a conclusion until the facts are sufficiently consolidated. This ability to live with uncertainty is not a temporary flaw; it is a structuring skill. It is a prerequisite for prudence, critical thinking and vigilance in the face of bias.
However, when the algorithmic tool immediately proposes a hierarchy of hypotheses or a plausible course of action, the experience of doubt can be shortened. Uncertainty is no longer experienced as a space to be explored, but as a gap to be filled quickly. Reasoning becomes validation rather than construction. The student confronts his or her thinking with an external suggestion even before having fully elaborated his or her own analysis.
This shift is subtle. It does not eliminate reasoning, but it modifies its temporality and intensity. The effort of hypothetication may diminish, not through incapacity, but through the habit of assistance.
Cognitive delegation cannot be decreed; it is established through an accumulation of micro-adjustments. Several mechanisms can be identified.
These mechanisms are not the result of an individual deficiency. They are linked to the very configuration of the system. As the philosophy of technology has shown, the technical environment shapes actions and expectations. Cognitive delegation is rarely conscious; it results from a gradual adaptation to the environment.
In regulated professions, clinical decisions entail full responsibility. The professional cannot hide behind a tool. Legally, morally and ethically, he or she remains the author of the decision.
However, the constant presence of algorithmic assistance can produce a paradoxical effect: responsibility remains formal, but the feeling of being the sole decision-maker diminishes. The decision appears to be co-produced by the system. In case of doubt, the "tool suggested it" argument may impose itself, at least internally, as a justification.
This phenomenon is in line with the ethics of responsibility developed by Hans Jonas: as technical power increases, so does the demand for responsibility. AI increases the capacity for analysis; it should therefore reinforce ethical vigilance. But if this power is accompanied by a subjective dilution of decision-making, the risk is the opposite: deciding without feeling fully in charge.
In training, this tension is crucial. Learning to practise a healthcare profession means integrating the fact that the final decision rests with oneself, in a context of irreducible uncertainty. If algorithmic assistance is experienced as an implicit guarantee, the construction of this responsibility can be weakened. Students risk associating the quality of their decision with compliance with the technical suggestion, rather than with the soundness of their reasoning.
Cognitive delegation is therefore not just a pedagogical issue; it is also an ethical one. It raises the question of how future professionals perceive themselves as responsible subjects in an algorithmic environment.
The risk of cognitive delegation does not mean we should give up on AI in training. Rather, it reveals a new requirement: to design devices capable of maintaining the effort of thinking, making reasoning explicit and reinforcing responsibility rather than attenuating it.
If assistance can become a crutch, it can also become a lever, provided it is integrated into a pedagogical architecture that is aware of its effects. The question is no longer just how to use AI, but how to train students in augmented reasoning without diminishing it.
It is therefore towards the design of demanding devices, articulating critical use and explicit metacognition, that we must now turn.
If AI reconfigures the conditions of clinical reasoning and exposes us to the risk of cognitive delegation, the response must be neither exclusion nor naive use. The pedagogical challenge is to transform assistance into a learning object. Training in augmented reasoning implies designing systems in which the tool does not shorten the effort of thinking, but rather puts it in tension, reveals it and demands it.
It's not just a question of teaching with AI, but of teaching the relationship with AI. This nuance is decisive.
A first pedagogical principle is to preserve spaces for unassisted reasoning. Students must be able to experiment with the autonomous construction of hypotheses, the progressive exploration of data and the assumption of uncertainty. AI must not be the reflex entry point to clinical cases.
In concrete terms, this means organizing contrasting sequences: initial analysis without assistance, formalization of hypotheses, then secondary confrontation with algorithmic proposals. This time lag is structuring. It enables the student to measure differences, identify convergences and understand divergences.
The aim is not to place the tool in competition with human reasoning, but to make visible the cognitive dynamics specific to each individual. AI thus becomes a critical mirror rather than a crutch. It reveals blind spots, questions intuitions, enriches the discussion without short-circuiting the initial elaboration.
This alternation also helps to maintain epistemic confidence. The student learns that he or she can produce a relevant analysis before any assistance. The tool then amplifies or questions this analysis, not replaces it.
Simply using a help system is not enough to train people to use it critically. We need to institute specific debriefing sessions focused not only on the clinical decision, but also on the interaction with the tool.
In these sequences, several questions can be explored:
The aim is twofold. On the one hand, to develop an algorithmic literacy adapted to regulated professions: to understand that AI works on probabilistic correlations and not on full contextual understanding. Secondly, to reinforce clinical metacognition: to be able to explain why one accepts, modifies or rejects a proposal.
This algorithmic debriefing transforms the tool into an object of analysis. It desacralizes technical suggestion without disqualifying it. It reminds us that all proposals, whether human or algorithmic, must be subjected to critical scrutiny.
Training in augmented reasoning also means rethinking evaluation methods. If a decision is assessed solely for its conformity to an expected response, the algorithmic tool can become an effective shortcut. On the other hand, if the assessment focuses on the quality of argumentation, the traceability of reasoning and the ability to explain uncertainties, the cognitive posture changes.
In a competency-based approach, the decision is only one element in the process. What matters is how it is constructed and justified. Integrating assessment criteria that focus on the explicitness of reasoning, on the identification of alternatives that have been discarded, and on consideration of the tool's limitations, helps to maintain intellectual standards.
It becomes relevant to assess the ability to question the AI itself: is the student able to formulate a relevant request? Detect inconsistencies? Contextualize the answer within the singularity of the patient? This critical dimension must be recognized as a professional skill in its own right.
Thus, augmented reasoning is not about thinking faster thanks to the tool, but thinking more lucidly with it. AI becomes a cognitive partner whose use requires discernment and responsibility.
Designing demanding systems is a necessary but not sufficient condition. Reasoned alternation, algorithmic debriefing and justification-based assessment all need to be part of a clear institutional framework.
Training in augmented reasoning means clarifying responsibilities, defining traceability rules and spelling out ethical expectations. The pedagogical requirement cannot remain implicit; it must be supported by shared principles and supervised systems.
In other words, the issue is no longer simply that of learning scenarios, but that of the governance of AI in training. It's this link between individual responsibility and institutional framework that we now need to focus on.
Training in augmented reasoning cannot rely solely on the quality of teaching scenarios. In the healthcare professions, decisions entail legal, deontological and ethical responsibilities that go beyond the simulation room. The integration of AI into training therefore calls for explicit institutional clarification. Without a framework, professional autonomy risks being silently eroded. With a structuring framework, on the contrary, it can be consolidated.
The challenge is not to control uses in order to restrict them, but to make them legible, debatable and accepted.
In the healthcare professions, responsibility for decision-making remains personal. No algorithm, however powerful, can be held legally responsible for a clinical act. This formal evidence must be reaffirmed in training. The tool may enlighten, suggest or alert, but it does not decide.
The digital environment tends to blur this distinction. When algorithmic suggestion is presented as highly reliable, the temptation is great to consider it as an implicit norm. The student may gradually equate compliance with the technical recommendation with a form of decision-making security.
The role of the institution is decisive here. It's up to them to remind people that AI is a tool to help them, not an authority. This means clarifying the place of these devices in training standards and regulations: neither forbidden nor prescribed, but integrated into a logic of critical use.
Clarifying "who decides" also means clarifying "who responds". Learners need to understand that ultimate responsibility lies with them, even when they have mobilized algorithmic assistance. This awareness should not be anxiety-inducing, but rather structuring. It is the foundation of professional identity.
One of the major transformations brought about by AI concerns the traceability of the decision-making process. If a tool has been used to develop a hypothesis or a course of action, this interaction must be made explicit.
In training, integrating this requirement means teaching students to document their approach:
The aim of this traceability practice is not to monitor, but to develop reflective awareness. It forces us to distinguish between what is a matter of personal analysis and what has been suggested by the system. It reinforces metacognition and prepares for transparent professional practice.
In a context where digital environments retain traces, the absence of explicitness can become problematic. Training to document the use of AI means anticipating future demands for quality and accountability.
To ensure that these principles do not remain theoretical, they must be enshrined in a shared framework. The development of a pedagogical charter for RNs in healthcare training is one way of achieving this.
Such a charter could be based on a number of guiding principles:
Here, the institution plays the role of guarantor. It must train its educational teams, define guidelines, and integrate these issues into school projects and skills frameworks. The absence of a framework leaves room for heterogeneous and sometimes contradictory practices, which undermine the coherence of training.
Protecting professional autonomy does not mean isolating future caregivers from contemporary technologies. It means creating an environment where the use of these technologies is thought through, discussed and regulated. Autonomy is not opposed to the framework; it is built into it.
Thus, the issue of AI in training is not limited to the performance of tools or the quality of pedagogical scenarios. It involves a complete architecture: individual responsibility, traceability of decisions, institutional governance.
But even the most rigorous framework does not dispense with personal accountability. Responsibility cannot be decreed; it must be embodied. Having examined the power of assistance, the risk of delegation and the conditions for institutional oversight, a more fundamental question remains: how can future professionals preserve the ability to doubt, to argue and to take full responsibility for their decisions in a world where algorithms are omnipresent?
It is to this question - that of the meaning and autonomy of judgment - that the conclusion of this article returns.
Artificial intelligence is neither an absolute rupture nor a simple technical continuity. It is part and parcel of the long history of the tools used to make medical decisions, but it is changing the intensity, speed and depth of those decisions. In training as in practice, it does not replace clinical judgment: it redefines its conditions.
This shift is decisive. The challenge is not to know whether AI is reliable - no human practice is perfectly reliable - but to understand what it does to the reasoning of those who use it. When suggestion precedes hypothesis, when the interface frames attention, when the answer is available even before the effort of elaboration, the relationship to doubt is transformed. Yet it is in confronting doubt that clinical competence is forged.
Training in the era of decision-support systems therefore presupposes a clear pedagogical choice: refuse to allow assistance to become the implicit matrix of judgment. This implies organizing the alternation between autonomous reasoning and algorithmic confrontation, instituting critical debriefing times, and evaluating the quality of argumentation rather than simply the conformity of the result. This also requires an explicit institutional framework, guaranteeing the primacy of professional judgment, traceability of usage and assumed responsibility.
Autonomy is not about making decisions without tools. It's about making informed decisions, taking responsibility for the technical mediations involved. In the healthcare professions, this autonomy is inseparable from responsibility. No algorithm will carry the moral burden of a clinical choice. The professional will.
Can we learn to decide if we no longer learn to doubt? This question runs through all our thinking. AI can enhance the quality of clinical reasoning, by revealing blind spots, broadening possibilities and stimulating discussion. But it can also accelerate the surface without deepening the structure. Apparent speed is no guarantee of inner solidity.
So it's not a question of giving in to technophile enthusiasm, or taking refuge in a posture of defiance. It's about recognizing that we are training professionals who will be working in an "algorithmized" environment, and that their skills will be not just technical, but reflexive. Their responsibility will not be diminished by the power of the tools, it will be enhanced by them.
In the final analysis, training in augmented reasoning means training to be doubly vigilant: vigilant in the face of human bias, but also vigilant in the face of algorithmic suggestions. It means learning to use AI without dissolving into it. It means maintaining the ability to suspend, to question, to argue, even when the answer seems already given.
In an assisted world, deciding remains a human act. And it is precisely this act that we, in training, must protect.
Illustration: AIg and training in the healthcare professions
Generated by AI (Canva) - Flavien Albarras
References
Agius, S., Magri, C., & Cassar, V. (2025). A Cognitive Task Analysis for Developing a Clinical Decision Support System for Emergency Triage. Journal of Emergency Nursing, 51(6), 1028-1045.e4. https://doi. org/10.1016/j.jen.2025.05.013
Aroulandom, F. (2025). Artificial intelligence and the nursing profession: Anticipating, understanding and accompanying an inevitable transformation. Sages-Femmes, 24(5, Supplement), S59-S60. https://doi. org/10.1016/j.sagf.2025.07.115
Auriault, P. (2014). Understanding medical imaging with Gilbert Simondon. Retrieved from https://philosophie.ac-amiens.fr/IMG/pdf/Simondon_Auriault-2.pdf
Académie Nationale de Médecine (2024). Generative AI systems in healthcare: Issues and perspectives (Report 24-03). Paris, France: Académie Nationale de Médecine. Retrieved from https://www.academie-medecine.fr/wp-content/uploads/2024/03/Rapport-Systemes-dIA-generative-en-sante.pdf
Bleher, H., & Braun, M. (2022). Diffused responsibility: Attributions of responsibility in the use of AI-driven clinical decision support systems. Ai and Ethics, 2(4), 747-761. https://doi. org/10.1007/s43681-022-00135-x
Digital Promise. (2024). AI literacy: A framework to understand, evaluate, and use emerging technology. Washington, DC: Digital Promise. https://digitalpromise. org/2024/06/18/ai-literacy-a-framework-to-understand-evaluate-and-use-emerging-technology/
Elsevier Masson (2023). Le raisonnement clinique infirmier. Paris, France: Elsevier Masson. Online presentation: https: //www.elsevier.com/fr-fr/connect/le-raisonnement-clinique-infirmier
Florey, S. (2023). Reconfiguration du littéraire par le numérique: Quelles potentialités didactiques? Revue de recherches en littératie médiatique multimodale, 18, 39-55. https://doi. org/10.7202/1108693ar
Gazquez-Garcia, J., Sánchez-Bocanegra, C. L., & Sevillano, J. L. (2025). AI in the Health Sector: Systematic Review of Key Skills for Future Health Professionals. JMIR Medical Education, 11(1), e58161. https://doi. org/10.2196/58161
Giroud, N. (2026). Use of AI in nursing education. La revue de l'infirmière, (317), 34-37. https://www. em-consulte.com/article/1788851/utilisation-de-l-ia-en-formation-infirmiere
Hoyelle-Pierre, S., & Jaillet, A. (2024). Learning clinical nursing reasoning through simulation in initial training for quality and efficiency of care. Revue internationale des technologies en pédagogie universitaire / International Journal of Technologies in Higher Education, 21(1), 37-54. https://doi. org/10.18162/ritpu-2024-v21n1-03
Inserm. (2024). Artificial intelligence and health. Paris, France: Inserm. https://www. inserm.fr/dossier/intelligence-artificielle-et-sante/
Izquierdo-Condoy, J. S., Arias-Intriago, M., Tello-De-la-Torre, A., Busch, F., & Ortiz-Prado, E. (2025). Generative Artificial Intelligence in Medical Education: Enhancing Critical Thinking or Undermining Cognitive Autonomy? Journal of Medical Internet Research, 27, e76340. https://doi.org/10.2196/76340
Montagna, M., Chiabrando, F., Lorenzo, R. D., Querini, P. R., & Students, M. (2025). Impact of Clinical Decision Support Systems on Medical Students' Case-Solving Performance: Comparison Study with a Focus Group. JMIR Medical Education, 11(1), e55709. https://doi. org/10.2196/55709
OECD. (2022). OECD Digital Education Outlook 2021: Pushing the boundaries with AI, blockchain and robots. OECD Digital Education Outlook, 2021. https://doi. org/10.1787/d5fe6bd0-fr
Position and recommendations of the Conseil national de l'Ordre des infirmiers regarding the use of artificial intelligence by nurses. (n.d.). Ordre national des infirmiers. Retrieved March 8, 2026, from https://www.ordre-infirmiers.fr/position-et-recommandations-du-conseil-national-de-l-ordre-des-infirmiers-relatives-a-l-utilisation
Patel, V. L. (2023). Reducing cognitive load in clinical decision support systems. HL7 Clinical Burden Working Group. Version PDF: https: //confluence.hl7.org/download/attachments/104568480/VLPatel-HL7%20WG--Clinical%20Burden-Dec%2011-2023-Final.pdf?version=1&modificationDate=1702318635272&api=v2
Sim, I., Gorman, P., Greenes, R. A., Haynes, R. B., Kaplan, B., Lehmann, H., & Tang, P. C. (2001). Clinical Decision Support Systems for the Practice of Evidence-based Medicine. Journal of the American Medical Informatics Association: JAMIA, 8(6), 527-534. https://doi. org/10.1136/jamia.2001.0080527
UNESCO. (2023). Guidance for generative AI in education and research. Paris, France: UNESCO. French version: https: //www.unesco.org/fr/digital-education/artificial-intelligence
Webster, C. S., Taylor, S., & Weller, J. M. (2021). Cognitive biases in diagnosis and decision making during anaesthesia and intensive care. BJA Education, 21(11), 420-425. https://doi. org/10.1016/j.bjae.2021.07.004
Zintchem, R., Ngono Zintchem, M. A., Ngae, D., Tembe-Fokunang, E. A., & Ntungwen Fokunang, C. (2025). Generative artificial intelligence and learning clinical nursing reasoning: Scope review, proposal of a "socioconnectrist" approach. Research and Advances in Nursing, (2). https://doi. org/10.25965/reasci.606
AI usage statement - ChatGPT and Perplexity were used as assistance tools for: (a) bibliographic review assistance (locating/sorting articles and structuring reading leads), (b) reformulation of certain passages to improve clarity and fluency, (c) spelling correction. AI did not produce arguments or data without validation: all references were checked and no quotations were invented. Content, analysis and interpretation remain my sole responsibility.
You might also like:
What if goals are a disservice to performance and learning?
Excessive focus on results can paradoxically undermine performance by generating anxiety and paralyzing action. Examples from sport and companies show that there are effective management and pedagogical alternatives.
Writing among teenagers, put to the test by digital technology and AI
Digital writing offers a lot more freedom, but its users also run many risks.
Assessing what's really valuable
Beyond rankings and sterile competitions, rethink school assessment to value the full range of skills developed by students.
Agritech: how to explain persistent famine?
the persistence of famine has its origins mainly in unfair policies, a food system held in an iron grip by powerful agribusinesses preoccupied with capitalist interests.
The library, our space for cultural transmission
The cultural transmission strategies of museums serve as an inspiration and many university and school libraries are taking the same path. They evolve into cultural spaces and assume in the best way the role of cultural transmission. An imposed culture wears like a scar, a favored culture shows itself as a pride.
