In July 2026, the TUMO Center for Creative Technologies published a strategy document that rethinks education, starting from one observation: artificial intelligence makes intellectual capability abundant, and what becomes scarce is authorship and connection. I offer here a complete summary, and a discussion drawing on my own practices of mediation, collective intelligence and cultural democracy.
A document born from listening to teenagers
Pegor Papazian, TUMO’s Chief Development Officer, sent me this document. We worked together on the AI monitoring committee of the Forum des images, which I led in 2024 and 2025 and of which he was a member; I reported on that long-term work in the article Webinar presenting the findings of the AI Monitoring Committee of the Forum des images. TUMO is a free out-of-school educational program, founded in Yerevan in 2011 and now present in more than twenty countries, where over 40,000 teenagers learn animation, game development, filmmaking, programming, music and design, with no grades, no exams, no diplomas.
The text, entitled Making, thinking and learning when capability is no longer scarce, is freely downloadable under a Creative Commons license. Its authors started from what teenagers themselves had to say, gathered at the ai/teens world conference of March 2025, designed and run by some 90 young people in sixteen cities. I told the story of that conference in the article What Teenagers Can Teach Us About Artificial Intelligence. One participant summed up his position in the vocabulary of video games: using AI is like using a cheat code, “but that doesn’t mean I let go of the controller.” Two concerns ran through those exchanges: keeping what is your own in your work, and protecting what is real in your ties to others. The TUMO document rests entirely on these two concerns.
Capability turned into a commodity
Throughout human history, the capability to make things (to code, to write, to produce images or sound) was scarce, and therefore valuable in itself. And producing anything worthwhile required other people: people to learn from, to build with, to think against. Work bound us together; belonging came as a by-product. With generative artificial intelligence, this capability becomes an abundant resource, a commodity. I developed my own analysis of this commoditization in the article Intelligence as a Resource; TUMO’s authors draw out its consequences for education.
When capability is abundant, two things become scarce. Authorship first: the part of the work that is our own, intention and judgment, since producing no longer requires mastering. Connection second: the ties that used to grow spontaneously out of needing other people no longer come of themselves, and must be deliberately built.
This shift touches three domains:
- Making (producing artifacts). When anyone can generate competent work on demand, value moves to what the person brings: deciding what is worth doing, bringing a point of view to it, directing and evaluating the result, answering for it.
- Thinking (reasoning, forming ideas). When fluent answers arrive instantly, what becomes scarce is judgment around reasoning: asking the right questions, telling what holds up from what merely sounds right, staying with a problem, keeping enough command of a domain to judge its outputs instead of submitting to them.
- Learning (building capabilities over time). When capability can be summoned instead of acquired, the traditional incentive to learn collapses. What remains essential is the patience mastery requires, the discernment that only knowledge of a domain provides, and the communities of practice that pass knowledge on.
Making, thinking and learning are the ground on which authorship and connection grow, and it is this ground that abundance erodes. A young person can now grow up able to summon any artifact without ever forging a voice of their own, able to obtain any answer without ever meeting anyone along the way.
Ten powers, from the person to society
The document is organized around a grid of ten “powers”, grouped into four sets ordered from the person toward society:
- Presence (turned toward oneself). Keeping command of one’s attention, through two powers: reflection (making room for wonder and exploration) and depth (staying with what is slow and difficult, committing over time).
- Agency (the engine of the whole). Contributing rather than consuming, through three powers: voice (originality, taste, a personal point of view, the courage to try something new), initiative (wanting to act, acting under uncertainty) and perseverance (seeing things through despite difficulty).
- Command. Knowing and understanding enough to work with authority and judgment, through three powers: knowledge (broad knowledge, curiosity, the pursuit of truth), skill (technique and tools, and the ability to learn new ones) and reasoning (analytical, critical and systemic thinking).
- Trust (turned toward society). Caring about the effects of one’s work on others, through two powers: responsibility (ethics, awareness of what is good for people) and sharing (communicating one’s contribution and its meaning so that others can take it up).
The authors are careful to say that these powers are not yet another competency framework. Each is precisely defined: sharing is not a general communication skill, but making one’s work transmissible (a silent learner who documents their work so well that a stranger could carry it on will score better on this power than a socially fluent learner whose work no one can pick up). Voice is not self-confidence, depth is not diligence. And these powers do not aggregate into a single score: each power requires a minimum level, below which the others degrade (voice without reflection becomes reflex, initiative without perseverance starts everything and finishes nothing), and above which each person develops a combination of their own.
Abundant capability, the authors write, does not diminish agency and command. By making execution cheap, it makes visible what was always the essential part of work: deciding what is worth pursuing, and turning it into something one can stand behind. If we delegate execution, the work can remain ours; if we delegate intention or judgment, we lose authorship.
Opportunities no longer come by themselves
The pedagogical part of the document rests on an idea that seems right to me. When capability was scarce, ordinary work provided by itself the opportunities to develop these powers: you had to rely on one another, weigh your decisions, wait, understand, make efforts, answer for what you did. With abundance, these opportunities disappear. Powers still develop the same way, through exercise, but the opportunities to exercise them must now be designed by educators, through the four components of a program: content (materials and tasks), pedagogical mechanics (the sequence of gestures asked of the learner), curriculum (the arrangement over time) and environment (people, spaces, interfaces).
Six design principles make it possible to restore, one by one, the lost opportunities, and TUMO’s practice illustrates each of them:
- Alternating working alone and working together. This principle restores the opportunity to rely on one another. At TUMO: individual self-learning side by side in shared spaces, collaborative workshops, and a deliberate absence of cohorts, with mixed ages and paths.
- Designing for autonomy, not compliance. This principle restores decision. At TUMO: each learner composes their own path through fourteen learning areas, and can leave at any time.
- Protecting time when nothing is due. This principle restores waiting, on which reflection and depth depend. At TUMO: no deadlines, no imposed duration, paths designed over years.
- Anchoring learning in questioning and judgment. This principle restores understanding. At TUMO: the “colearner”, a virtual peer who learns alongside the learner, sometimes showing, sometimes asking to be taught, which meets the create-rather-than-explain pedagogy I defend in the article Pedagogy of Artificial Intelligence.
- Making difficulty safe. This principle restores effort. At TUMO: no grades, no exams, no diplomas; a failure costs only the time it takes to start again; the portfolio grows with the projects and serves as a “living diploma”.
- Treating learners as creators, not trainees. This principle puts each learner back in the position of answering for their work. At TUMO: projects published, submitted to festivals, presented in public by their authors.
The authors finally propose holding two time horizons together. The jagged present, in which AI excels only in some domains, and where the fundamental skills it has not superseded must therefore still be developed. And a maximalist future, in which it would do most complex intellectual tasks better than most humans, to be considered both as a thought experiment and as a scenario that can no longer be ruled out.
The school of opportunities versus the school of delivery
The chapter I find strongest is about school. One might think that the perfect AI tutor, that old dream of education research since Benjamin Bloom’s work on individual tutoring (1984), makes school obsolete. TUMO’s teams answer with a formula of their own: “AI can teach you what you don’t need to learn.” If a machine can teach you a skill, it can also perform it for you, and the incentive to master it weakens. Yet the tutor perfects instruction, which is precisely the part of school that is losing its value. The powers, for their part, cannot be delegated; they develop only through exercise, in opportunities that school is the only institution able to guarantee for everyone. “When capability becomes a commodity, school does not become less necessary. It becomes more important, but it becomes something else.”
I read in this reversal an answer, half a century later, to Ivan Illich, who in Deschooling Society (1971) attacked school as a monopoly on instruction and proposed to deinstitutionalize it. The authors take up the critique of instruction, but draw the opposite conclusion: now that instruction is available everywhere, school recovers its oldest raison d’être, a protected time and space that guarantee everyone the conditions of their development. They name the equality issue without evasion: well-off families already organize these opportunities at home (protected time, mentors, a room turned into a workshop, an audience for first attempts). If opportunities become a private good, the divide that used to run through access to instruction reappears, deeper, in access to the conditions for developing authorship and connection.
The educator changes along with the school. Freed from the role of main source of capability, they occupy the positions only a person can hold: the witness whose gaze matters (risking something of your own takes courage only in front of someone who could judge you), the person one answers to (an expectation held by someone can be discussed and disappointed in a way no specification allows), and the author of the opportunities themselves.
The authors finally contrast two possible schools, built with the same tools. The school of opportunities, just described. And the school of delivery, where abundance serves to intensify transmission: content generated at machine speed, automatic grading at volume, personalization pushed to the point of isolating each student in their own stream. Each “efficiency” of this second school costs an opportunity. The difference between the two lies not in equipment but in design, and the six principles serve as a practical test for telling them apart. The authors acknowledge that the school of delivery is the easier one to build, and even the more rational one under current accountability regimes: the philosopher C. Thi Nguyen calls value capture the mechanism by which an institution adopts a simplified, countable version of what it values, until the simplified version takes the place of the value itself. Getting out of this does not depend on schools alone, but on public policy: making room in school time, grounding accountability in instruments that measure the powers, admitting the portfolio alongside the exam.
Assessing the program, not the person
The last chapter addresses a difficulty I meet in all the organizations I work with: how to assess, when work can be produced without the capabilities it used to attest, and a fluent account of the process can be produced just as easily? Banning the tools solves nothing, since you then measure a performance under conditions the learner will never encounter again. The authors propose a reversal: making abundance the source of the evidence. Situations are staged in which automated capability is within easy reach (a ready-made solution offered at the very start of the work, a change of brief midway, a seductive but ill-suited option in work that will be published), and it is the learner’s choices that reveal the powers. Heavy use of the assistant is not penalized, as long as the idea remains the learner’s.
The main instrument is a series of three challenges, one per domain:
- The making challenge. A small commission for a client, for instance a poster for a local association. Powers observed: voice, perseverance, responsibility.
- The thinking challenge. An inquiry into an open question, based on documents of uneven quality, including one confident but flawed analysis, with a final judgment to communicate to a specific recipient. Powers observed: knowledge, reasoning, sharing.
- The learning challenge. Getting to grips with an unknown tool, producing something with it, writing a guide another person could follow. Powers observed: initiative, skill, depth, sharing.
Reflection is measured in all three challenges by the same device: a first idea recorded, an imposed pause of a few days, the idea recorded again; the gap between the two versions serves as the measure. The scenario is constructed (the client is fictional, the interactions are scripted so that everyone faces the same situations), but the result is real: the finished work goes into the portfolio, sometimes handed over to a real commissioning organization. And the grading criteria describe how the learner worked, not the polish of what they produced.
This measurement targets the program, not the individual, and to my mind this is the most important point. Results are aggregated by cohort, no one is graded or ranked, and this choice is consistent with the substance of the approach, since each learner is expected to arrive at their own combination of powers. The legitimate question is not to rank people, but to know whether the program develops the full range of powers. In the article The question of criteria I raised the question of what criteria are for at school; in this device I find an answer that can be put to work, one that shifts assessment from the person to the institution.
Five meeting points with my own practices
This document touches me because I find in it, carried by a strategy, arguments and tools, intuitions I have been working on for years in mediation, collective intelligence and cultural democracy.
The first is the stance of the capable person. The sixth principle, treating learners as creators and not trainees, is the foundation of any mediation worthy of the name: considering every person capable from the outset, and building the framework that lets them exercise that capability, which I described in the article The framework that authorizes. It is the postulate of the equality of intelligences, and TUMO’s whole design follows from it: if you believe young people are capable, then you entrust them with real commissions and real responsibilities, in front of real audiences.
The second is putting the process before the result. In my workshops, what matters, and what I report back, is the process people have lived through, never the mere quality of the objects produced. With its assessment design, TUMO institutionalizes this shift: in an era when the result can be generated, only the process still attests to what a person is developing. What mediators knew by intuition becomes a necessity of method for everyone.
The third is collective intelligence. The question I keep asking myself in my facilitation work, and which I worked on in the article Leading Collective Intelligence in the Face of Crises, is this: under what conditions does collective intelligence actually emerge when people are brought together? TUMO’s authors shed light on this question by showing that connection, once a by-product of work, becomes an object of design: the alternation between working alone and working together, the mixing of ages, the peers and the real recipients are choices of social architecture, not givens. When cooperation is no longer technically necessary, it must be made desirable and deliberately built, and that is the very craft of collective intelligence facilitators.
The fourth is the link with psychosocial skills, which I develop through artistic practices in my training courses, as I recounted in the article Developing Psychosocial Skills Through Artistic Practices. The ten powers partly overlap this field (responsibility, sharing, perseverance) while explicitly setting themselves apart from it. The authors insist on this point: their definitions cannot be reduced to existing social-emotional frameworks, because the commoditization of capability has shifted their meaning. This articulation can be of use to French psychosocial-skills practitioners: it gives a new anthropological foundation to what they already do.
Finally, cultural democracy. TUMO’s program rests on a question asked by its founders, that of a pedagogy of the possible departure: what would a school be like that you could leave at any moment? Popular-education movements have been answering this question in practice for a century, since no one is ever obliged to come to their workshops. The idea that each person arrives at a combination of powers of their own, irreducible to a score, meets the affirmation, central to cultural rights, of the dignity and singularity of each person. And when the authors conclude that the non-formal will be “mainstreamed”, because what it has always aimed at directly (the powers at work within the work, rather than the work itself) becomes what school will have to deliver, they place non-formal education in the position of a laboratory: it is there that the evidence that will change the rules can first be established. For the culture and education practitioners I work with, this is a reversal of legitimacy, and a political argument to seize.
I remain vigilant on one point. The authors write it themselves: their powers are not a framework; but institutions always end up wanting to apply a grid to individuals, and the capture mechanism described by Nguyen could apply to the powers as it applied to grades. The protection lies in the choice, reaffirmed throughout the document, never to grade or rank people; that choice is political, and it will have to be defended.
The making of the document as a demonstration
One last detail gives the text its coherence. In the section “How this document was made”, the authors state that they wrote the text with artificial intelligence tools throughout; some passages started as machine drafts, later reworked, and no claim entered the text “on the sole authority of a model”: the questions, the concepts, the principles and the conclusions remain the team’s. The division of labor described is the very one the authors theorize: execution partly delegated, intention, judgment and responsibility retained. It is also mine, which I described in the article The skill as a work of art: this document on authorship in the age of abundance is itself an exhibit in the file of what an author is becoming.
