(Included at the end of this manifesto is a free self-assessment tool to evaluate your readiness for AI and help you move toward a more critical, responsible, and human-centered approach to its use)
By Dr. Tawhid CHTIOUI, Founding Presidentof aivancity School of AI & Data for Business & Society; selected by Keyrus as one of the 25 most influential global figures in the field of AI and data (January 2025).
Introduction: The Night ChatGPT Replaced the Father
The other night, I experienced a special moment—one of those tiny moments that at first seem to belong to everyday life, but which suddenly reveal a much deeper transformation in our relationship with knowledge.
My son was working on a homework assignment. I approached him with that quiet confidence typical of parents who still believe that their mere presence can clarify a question, resolve a problem, or save the day. I asked him, with all the paternal dignity I could muster at that hour: “Do you want me to help you? ”
He looked up at me and replied very simply, “ No, Dad, it’s okay. I already asked ChatGPT.”
I could have smiled and let it slide. But I instinctively tried to salvage what was left of my authority as a parent. So I asked him, “ So, what did he say? ” His reply was almost devoid of malice, which made it all the more effective: “ He explained it to me better than you did last time.”
At that moment, it would have been easy to file this scene away in the family’s vast collection of children’s little acts of defiance. Yet something held me back. This remark was not merely an amusing anecdote. It encapsulated, in the simplicity of an ordinary scene, a fundamental shift in human behavior.
If a child can sincerely believe that a machine explains things better than a parent, clarifies them more effectively, rephrases them more patiently, and responds more efficiently, then the issue is no longer merely technical. It is no longer even merely educational. It touches on our very definition of intelligence.
For centuries, we have placed intelligence on a special pedestal. We have admired those who knew the answers, those who answered quickly, those who found the right solution before others, those who calculated without hesitation, memorized without faltering, and argued without wavering. Our schools have largely embraced this vision, our competitive exams have enshrined it, our tests have timed it, and our companies have rewarded it. A large part of our social systems has been built around an implicit equation: to be intelligent is to produce the right answer, at the right time, faster or better than others.
This idea was not absurd. It even had its moments of greatness. It shaped generations of doctors, engineers, teachers, researchers, leaders, and builders. It made it possible to structure knowledge, identify talent, organize the transmission of knowledge, and place value on effort, memory, rigor, and proof, but today it is no longer enough.
Generative artificial intelligence is now capable of performing a significant portion of these tasks. It writes, summarizes, translates, compares, corrects, codes, simulates, proposes outlines, explores hypotheses, and rephrases arguments. It can explain a difficult concept in several different styles, start over without getting impatient, rephrase without tiring, adapt its level without taking offense, and never imply—as we sometimes do despite ourselves—that the question has already been asked too many times.
That is precisely what troubles us. AI does not impress us merely because it calculates quickly or produces a great deal. It unsettles us because it is encroaching on activities we once considered deeply human: explaining, writing, advising, conversing, and guiding. It is no longer content to automate peripheral tasks. It is entering the sensitive realm of speech, reasoning, and knowledge transfer.
As early as 1950, Alan Turing posed a question that has since become seminal: Can a machine mimic intelligence to the point where we can no longer distinguish its behavior from that of a human? For a long time, this question was confined to laboratories, philosophers, and computer scientists. It has now entered our living rooms, classrooms, offices, and phones. It arises when a child prefers to ask ChatGPT rather than their father.
The real shock, however, is not that the machine provides answers. The real shock is that we have for so long reduced intelligence to that single ability.
If intelligence were merely the speed at which we arrive at the right answer, then we would have to admit, bluntly, that we are already outmatched. An AI can read faster than we can, synthesize information faster than we can, compare faster than we can, produce results faster than we can, and explore in a matter of seconds options that would take us hours to formulate. But this conclusion would be too simplistic, because it would confuse cognitive performance with human intelligence, the production of an answer with the act of thinking, efficiency with understanding, and speed with depth.
Herbert Simon, winner of the Nobel Prize in Economics, demonstrated that our rationality is never absolute. It is limited, context-dependent, and constrained by time, available information, contexts, and objectives. Daniel Kahneman and Amos Tversky, two leading figures in cognitive psychology and behavioral economics, have, for their part, revealed just how much our judgments are influenced by biases, mental shortcuts, and sometimes misleading intuitions. In other words, human intelligence has never been a mere machine for producing exact answers. It has always been a way of making decisions amid uncertainty, with incomplete information, emotions, values, responsibilities, and consequences.
This is where AI forces us to make a decisive shift. It does not spell the end of human intelligence. It spells the end of a definition of intelligence that is too narrow.
We aren’t being outpaced by AI; we’re being outpaced by our old way of defining what it means to be intelligent.
The distinction is crucial. If we continue to measure human intelligence using the very criteria on which machines are becoming unbeatable, we will be setting the stage for our own symbolic decline. We will create generations who see themselves as slow, fragile, and imperfect versions of artificial systems. We will be asking humans to compete with machines on the machines’ own terms. That would be a historic mistake.
A human being is not just a slower AI.
It is not a biological processor to which emotions, fatigue, and a few hesitations have been added. It is not an imperfect machine with the flaw of doubting, forgetting, loving, fearing, and making mistakes. It is something else. And it is precisely this “something else” that we must now name.
Because AI forces us to ask a question we should have asked long ago: What is intelligence when answers are no longer scarce? In a world where answers are becoming abundant, instantaneous, and sometimes brilliant, can we still define intelligence as merely the ability to provide answers?
This is where the real debate begins. Not: What can AI do? But rather: What does it reveal about the way we think, learn, work, and make decisions?
If machines can generate answers, what will schools still need to teach? If they can write, analyze, compare, and make recommendations, what will the professional world still need to value? If they become part of our choices, our careers, our institutions, and our everyday lives, what new divide might we see emerging?
This is the path I propose we explore: not AI versus humans, but AI as a catalyst for a deeper, older, and more decisive question: what do we now mean by intelligence?
I. Redefining Intelligence: The Five Dimensions of Human Intelligence in the Age of AI
1- The Art of Asking Questions
The first aspect of this intelligence that needs to be rebuilt is undoubtedly the oldest, but also the most neglected: the art of asking the right questions.
We still live with the notion that a question is merely a step toward the answer, a sort of temporary void destined to vanish as soon as knowledge arrives. Yet, in a world saturated with answers, the question becomes a decisive act. It is no longer a sign of ignorance. It becomes the first form of judgment.
AI provides the answers. Humans ask the questions.
This phrase may seem simple, but it signals a revolution. Asking a question isn’t just about seeking information; it’s about choosing a perspective, deciding what’s worth examining, distinguishing the symptom from the problem, the urgent from the essential, and superficial curiosity from genuine inquiry.
A student who asks an AI to “give a presentation on climate change” isn’t engaging in the same intellectual effort as one who asks, “How can we explain that societies have been aware of climate risks for decades yet continue to delay collective action?” In the first case, the machine will generate content. In the second, it will be drawn into a problem. The difference is immense. The first request calls for a summary. The second opens an investigation.
John Dewey, the American philosopher and pragmatist educator, noted that learning truly begins when a situation becomes problematic—when it challenges our assumptions and compels us to investigate. This idea is strikingly relevant today. In the age of AI, learning is no longer just about receiving an answer, but about knowing how to frame the problem that warrants an answer.
In a world where information was scarce, intelligence often meant knowing where to look. In a world where answers are plentiful, it means first and foremost knowing what to ask—and, above all, knowing why we ask it.
This is where the first divide emerges. Those who ask poor questions will receive poor answers, even if they are well-written. Those who ask confusing questions will receive answers that seem clear but are based on a misunderstanding. Those who ask closed-ended questions will get effective but narrow answers. The power of the tool will not compensate for the poor quality of the questions; it may even amplify it.
2. The Wisdom of Discernment
The second dimension is discernment.
Because AI doesn’t just generate answers. It often generates plausible results. And plausibility has become one of the most dangerous things of our time.
An answer can be fluid, coherent, elegant, and convincing, yet still inaccurate. It can take on the appearance of truth without possessing its substance. It can give the impression of understanding when it is merely restating linguistic patterns.
Emily Bender and Timnit Gebru, researchers in computational linguistics and AI ethics, have popularized the term “stochastic parrots” to highlight the fact that large language models manipulate statistical patterns without having any firsthand experience of the world they describe.
The statement has sometimes been debated and challenged, but it does serve to remind us that a machine can produce a very convincing text without guaranteeing that the text is true, accurate, or complete.
That is why discernment has become a key skill. It is no longer just a matter of knowing how to read. One must know how to question, verify, contextualize, compare, and prioritize. One must learn to hear what is missing from an answer as much as what is included in it. One must identify the missing source, the unspoken bias, the implicit assumption, and the dominant viewpoint that presents itself as neutral.
We are no longer merely in a knowledge economy. We are entering an economy of discernment.
For a long time, mistakes were often caused by a lack of information. In the future, they will increasingly stem from placing too much trust too quickly in well-presented information. Ignorance had at least one virtue: it sometimes knew that it was ignorant. False knowledge, on the other hand, marches forward with confidence.
Yet human intelligence begins precisely where the possibility of doubt arises. Being able to say “I’m not sure” is not a weakness. It is an achievement. It is what distinguishes critical thinking from the mere consumption of answers. It is what prevents fluidity from turning into hypnosis.
3- The Intelligence of Meaning
The third dimension is the ability to give meaning.
AI can detect correlations, identify patterns, categorize preferences, anticipate behavior, and optimize processes. It can tell us what is likely, common, effective, or statistically consistent, but a correlation is not an end in itself. A prediction is not a direction. Optimization is not a project.
The machine can help choose the means. It cannot, on its own, determine the end.
This distinction is fundamental. In a company, AI can suggest ways to maximize a team’s productivity, but it cannot single-handedly decide what social model that productivity should serve. In a city, it can optimize traffic flow, but it cannot single-handedly decide how much importance we want to place on slowness, social interaction, and quality of life. In education, it can personalize exercises, yet it cannot single-handedly define what it means to educate a free, cultured, and responsible citizen.
Edgar Morin has taught us that understanding does not consist of isolating fragments, but of connecting them into a living whole, permeated by uncertainty, contradictions, and purposes. This is exactly what AI is forcing us to relearn. We need powerful tools to process information, but we need even more people who are capable of transforming that information into guidance.
Giving meaning means refusing to confuse what is possible with what is desirable. It means remembering that an innovation is not good simply because it is new, that a decision is not right simply because it is optimized, and that a solution is not humane simply because it is effective.
4. Connective Intelligence
The fourth dimension is connective intelligence—that is, the ability to bridge worlds that have been viewed as separate for far too long.
Major contemporary issues do not respect disciplinary boundaries. AI in healthcare is not merely a medical or computer science issue; it involves law, ethics, patient relations, trust in institutions, work organization, and data protection. AI in finance is not merely a matter of computation; it raises questions about risk, accountability, inclusion, and transparency. AI in education is not merely a pedagogical issue; it touches on equal opportunity, teacher training, citizenship, and the relationship to truth.
The intelligence of tomorrow will be less about accumulation and more about connection. It will not consist merely of piling up knowledge, but of circulating meaning between different bodies of knowledge.
This is a significant shift. For a long time, we have valued experts capable of achieving great depth in a single field. Such expertise will remain essential. But it will no longer be enough. The world will need professionals capable of bridging boundaries, translating between different fields, and facilitating dialogue between engineers and lawyers, doctors and data scientists, educators and designers, and leaders and citizens.
The complexity of the world cannot be solved by isolated minds. It will require minds capable of making connections.
5. Responsible Intelligence
Finally, the fifth dimension is responsibility.
This may be the aspect that most profoundly distinguishes human intelligence from artificial intelligence. An AI can suggest, categorize, recommend, generate, and automate; nevertheless, it is not accountable to society. It does not look a patient in the eye. It does not explain to an employee why a decision led to their dismissal. It does not apologize to a student whom a system may have unfairly penalized. It does not grasp the gravity of a consequence.
Humans, on the other hand, cannot hide behind the machines they use. This statement should become a central ethical principle of our time.
Delegating a task does not mean delegating responsibility. Automating a decision does not mean absolving the person who conceived, approved, or allowed it to proceed. The more powerful systems become, the more explicit human responsibility must be.
Hannah Arendt reminded us that human beings are also defined by their ability to take responsibility for their actions in a shared world. This idea becomes crucial in the age of automated systems. Human intelligence is not merely cognitive; it is moral, because it entails consequences for others beyond oneself.
That is why the contrast between artificial intelligence and human intelligence is a false dichotomy. The real question is not which of the two should prevail. The real question is how to facilitate their cooperation without erasing what makes humans accountable.
We are entering an era in which we will increasingly think alongside machines. But thinking alongside a machine must never mean thinking in its place, nor letting it think for us. This requires a new inner discipline: knowing how to question the tool, verify its outputs, understand its limitations, take responsibility for its uses, and keep alive what the pace of technological advancement sometimes tends to overwhelm—our judgment.
AI processes information. Humans inhabit the world. AI calculates possibilities. Humans choose a direction. AI produces a response. Humans must take responsibility for that response.
Perhaps this is where the new definition of intelligence begins to take shape: no longer merely providing the correct answer, but asking the right question, distinguishing the true from the plausible, connecting facts to a purpose, bringing different fields of knowledge into dialogue, and accepting the consequences of one’s decisions.
This redefinition is not abstract. It will transform schools, universities, professions, organizations, management, and citizenship. For when intelligence shifts its center of gravity, all institutions that claim to shape, assess, or utilize it must, in turn, transform themselves.
II. What This Means for Education: From a School of Reaction to a School of Discernment
If intelligence is no longer defined primarily by the ability to produce an answer, then education can no longer be organized as if that answer were its primary goal.
This is a profound shift, because the modern school system has largely been built around a simple agreement: the institution imparts knowledge, the student learns it, and exams verify that the student can recall or apply it. This pact was coherent for a long time. It made it possible to structure learning, organize progression, ensure a common foundation, recognize effort, and certify skills.
However, the advent of generative AI is throwing this model into turmoil. Not because knowledge is becoming obsolete, but because the way we access, formulate, and apply it is changing. Students are no longer alone facing a blank page. They are no longer alone facing a question. They are no longer alone facing ignorance. They have, at their fingertips, an assistant capable of suggesting, rephrasing, explaining, correcting, and producing.
Therefore, continuing to assess as if this assistant did not exist amounts to staging a pedagogical charade.
This illusion may offer some reassurance for a time. It allows us to maintain the appearance of a stable world, one in which a student’s own work is clearly distinct from outside assistance, one in which a student’s response can always be identified as the pure product of their own mind, and one in which a teacher can still unambiguously distinguish between what is learned and what is delegated. That world is no longer ours.
This does not mean that we should conclude that everything should be permitted, nor that individual effort is a thing of the past. That would be an equally serious mistake.
Education should neither idolize AI nor demonize it. It should help guide and shape it.
The question, then, is not simply: should we allow or ban ChatGPT, Gemini, or Claude in homework assignments, exams, or theses? While this question is necessary, it is not sufficient. The real question is more challenging: what do we still want to assess when machines can generate an increasing proportion of the answers we once asked students to provide?
This is where schools must shift their focus. It must shift from a pedagogy of reproduction to a pedagogy of discernment. It must stop viewing the final answer as the sole measure of intelligence, and instead focus more on the path that leads to it: the quality of the question asked, the relevance of the sources used, and the way in which the student has verified, corrected, challenged, improved, and taken ownership of what they produce.
School or college work can no longer be merely a final product to be submitted. It must become a line of reasoning that is made visible.
We should no longer simply ask a student, “What is your answer?” We should also ask them: “How did you arrive at that answer? What did you ask the AI? Why did you phrase your question that way? Which answers did you reject? What errors did you spot? Which sources did you verify? What biases did you identify? How much of the decision do you personally take responsibility for?”
This shift is essential. It transforms AI from a tool for cutting corners into a tool for setting high standards. Used without a clear framework, it can become a means of avoiding intellectual effort. When integrated intelligently, it can become a means of enhancing intellectual effort.
This distinction is crucial. Because the effort does not disappear; it shifts. Yesterday, the effort often consisted of producing an answer on one’s own within a limited time frame. Tomorrow, it will also involve navigating a complex cognitive environment, consulting multiple sources, comparing hypotheses, identifying errors, documenting one’s choices, distinguishing between what comes from oneself and what comes from the tool, and transforming an automated output into personal thought.
It’s not any less demanding. It’s more demanding.
Lev Vygotsky, a developmental and learning psychologist, demonstrated that intelligence is always constructed through mediators: language, symbols, tools, and social interactions. AI can be understood as a new form of cognitive mediation, one of unprecedented power. It is not merely an external tool that we use; it alters the very way we search, formulate, learn, and think. However, all forms of mediation must be learned.
A tool that enhances can also weaken. A crutch can help someone walk, but it can also prevent them from relearning how to stand on their own if it becomes a permanent aid. The challenge in education, therefore, is to ensure that AI is not a crutch for laziness, but a partner in progress.
This requires rethinking assignments. Assigning a traditional essay to be completed at home, without allowing for the use of AI, is becoming problematic. On the other hand, asking students to produce a first draft using AI, then to analyze it, critique it, expand upon it, verify its sources, identify its limitations, and justify their choices can become a much more educational exercise. The challenge is no longer to hide the tool, but to ensure its use is intelligent, traceable, and responsible.
This also means rethinking exams. We will need to maintain opportunities without AI, because certain fundamental skills must be internalized. We cannot form sound judgments if we outsource all memory, calculation, writing, or reasoning too early. At the same time, we must also create opportunities involving AI, because it is in this environment that students will live, work, and make decisions. Training them solely in a world without AI would be as artificial as training pilots without simulators or doctors without imaging.
The question isn’t whether to learn “with” or “without” AI. The question is what each situation allows us to learn.
Without AI, we can assess personal mastery, deep understanding, and the ability to reason without immediate assistance. With AI, we can evaluate the ability to steer the process, maintain a critical perspective, engage in dialogue with a system, refine one’s work, and exercise responsibility in an augmented environment. Both are necessary. What would be dangerous is to confuse one with the other.
In this new educational landscape, the teacher’s role does not disappear. It becomes more nuanced, more strategic, and even more human.
The teacher is no longer merely the one who possesses an answer that the student does not. Instead, the teacher becomes the one who helps create the conditions for sound thinking. They learn to formulate a problem, to not settle for a tempting answer, to identify a weak source, to understand an error, to situate an idea within a historical context, and to connect knowledge to a consequence. They become less a distributor of content and more an architect of discernment.
Philippe Meirieu, a French educational theorist, points out that education is not simply about imparting knowledge, but about enabling individuals to develop themselves through a challenging engagement with the world. This idea takes on central importance in the age of AI. The danger is not that students will use machines to produce assignments. The danger is that they will become accustomed to producing work without developing themselves.
A machine can explain a concept—and do so very well—but it doesn’t always know why a student doesn’t understand. It doesn’t necessarily see the shame behind the silence, the fear behind the sarcasm, or the discouragement behind the lack of questions. It doesn’t know what it means, in the course of a person’s life, to receive at just the right moment a remark that lifts one up rather than corrects them.
Education is not merely the dissemination of standardized answers. It is a transformative relationship.
That is why teachers must remain at the center—not as nostalgic guardians of a bygone era, but as clear-sighted mediators of the new world. Their mission will not be to compete with AI in its ability to provide instant explanations.
The teacher’s role will be to teach students not to confuse explanation with understanding, assistance with independence, or fluency with truth.
This shift must also transform degrees. Too often, we continue to view them as certificates of the past: at a given point in time, an institution certifies that a person possesses a set of knowledge and skills. But in a world where knowledge, professions, and tools are evolving rapidly, this snapshot quickly becomes outdated.
The diploma of the future should be less of a monument and more of a living organism. It will no longer be enough for it to simply say, “This person has learned.” It will also have to say, “This person knows how to keep learning.”
He must demonstrate the ability to stay current, adopt new tools, understand new contexts, and maintain sound judgment in changing environments.
This is likely one of the major transformations in higher education. Education will no longer consist solely of preparing students for their first job. It will need to prepare them for multiple careers, for professions that will evolve, and for skills that will need to be regularly reassessed. School can no longer be merely a place one enters before entering the workforce. It must become a lifelong companion to the mind.
This development also has a democratic dimension. While AI can serve as a catalyst for learning for those who know how to use it, it can also exacerbate inequality for those who lack the right tools, support, and critical frameworks.
Some students will learn to think with AI. Others will simply ask it to do the work for them. The former will gain intellectual power. The latter risk losing their independence.
This is why the integration of AI into education cannot be left solely to individual use. It must become an institutional, educational, ethical, and social initiative. We will need to train teachers, clarify guidelines, adapt assessments, support students, and foster a culture of critical thinking, source verification, and accountability. We must also ensure that access to these tools does not further widen existing divides.
The goal is not to create a school obsessed with technology. The goal is to build a school capable of fostering free-thinking minds in a world saturated with technology.
The purpose of education has never been merely to impart skills. It is to shape individuals capable of navigating the world with clarity—people who can understand before acting, question before believing, connect the dots before deciding, and take responsibility for their actions.
AI doesn’t make this mission less important. It makes it urgent.
III. What This Means for the Professional World: From Employee-Executor to Professional-Orchestrator
If the definition of intelligence changes, work cannot continue to be organized around the same criteria for value.
For a long time, the professional world, too, valued responsiveness. The ideal employee was often someone who could produce results quickly, handle a case, draft a memo, prepare a presentation, analyze a spreadsheet, respond to a request, and carry out a task reliably. It was important to know how to do things, but above all, to do them on time. Competence was largely measured by the ability to turn instructions into deliverables.
This approach will not disappear entirely. Organizations will always need discipline, reliability, execution, and technical expertise; however, AI will gradually shift the boundary between what is the domain of human production and what can be automated, assisted, or augmented.
In many professions, a significant portion of routine intellectual work is already becoming automatable: summarizing a document, drafting an email, producing an initial analysis, comparing quotes, generating code, preparing a legal summary, formulating a business recommendation, segmenting data, translating content, building a sales pitch, and designing meeting materials. These are tasks that used to take up a considerable portion of work time. They are not disappearing, but their status is changing. They are becoming less often the core of value and more often the starting point for more demanding work.
The professional of tomorrow will therefore not simply be someone who produces faster than a machine. They will be someone who knows how to extract useful output from the machine, interpret it, correct it, enhance it, put it into context, and use it to make responsible decisions.
In other words, work is no longer just about carrying out tasks. It is increasingly about coordinating.
This orchestration is not a vague or superficial skill. It requires a deep understanding of the issues, tools, constraints, risks, and objectives involved. Simply asking an AI to do something is not enough. One must also know what to ask it to do, within what framework, using what data, with what limitations, for what purpose, with what level of verification, and under whose responsibility.
In this new economy, the most valuable professional won’t necessarily be the one who can do everything on their own. It will be the one who knows how to work with multiple forms of intelligence without losing their judgment. They will be able to articulate a business problem, choose the right tools, test several hypotheses, compare results, identify errors, translate technical output into understandable decisions, and, above all, assess the human, economic, legal, or ethical consequences of what they propose.
Peter Drucker, a pioneer of modern management thought, popularized the concept of the “knowledge worker”—a professional whose value lay in their ability to harness information and expertise. AI takes us a step further: into the era of the augmented professional—not because they possess more information, but because they can navigate a cognitive environment composed of humans, data, models, platforms, and intelligent agents.
This shift is fundamentally changing what companies expect. They will no longer be looking solely for candidates who can follow a procedure or use a tool at a given moment. Instead, they will be seeking people who can learn quickly, understand concepts in depth, work effectively with technical systems, and maintain sound judgment in uncertain environments.
The focus will shift to five skills that directly align with my proposed new definition of intelligence.
First,the ability to frame problems. In organizations, many failures stem not from a bad solution, but from a bad problem. AI can optimize a response, but it does not guarantee that the question being asked is the right one. The professional we need is someone who knows how to reframe a request, identify the real issue behind the symptom, resist overly quick fixes, and transform an operational challenge into a clear problem.
Next, the ability to discern. The professional world will be inundated with generated content: reports, analyses, forecasts, recommendations, minutes, scenarios, and dashboards. Everything will seem smoother, faster, and more presentable. Not everything, however, will be reliable. Competent employees will need to know how to verify, source, compare, and identify biases; distinguish robust data from fragile indicators; and tell relevant recommendations from appealing conclusions. In an AI-enhanced organization, discernment will become a safety skill.
Next comes the ability to make sense of things. Companies will have no shortage of tools for measuring, forecasting, and optimizing. What they will sometimes lack are people capable of linking these metrics to a strategy, a culture, a customer promise, or a shared vision. AI can help identify what works. It cannot decide on its own what is worth pursuing.
A valuable professional is someone who knows how to turn analysis into direction, data into decisions, and performance into progress.
The fourth skill will be the ability to bridge worlds. Organizations already suffer from silos. AI can reinforce these silos if it remains confined to technical applications, or break them down if it becomes a common language across business units. Companies will need translators, facilitators, and architects of collaboration—people capable of fostering dialogue among senior management, business teams, the IT department, legal, human resources, customers, partners, and sometimes regulatory authorities.
Ultimately, the most critical factor will be accountability. The more decisions are supported by AI, the more important it will be to know who is actually making the decisions, who is approving them, who is overseeing them, and who is taking responsibility for them.
An algorithmic recommendation must never become an organizational excuse. “The tool said so” cannot become the new form of managerial abdication of responsibility. Delegating part of the processing does not mean delegating responsibility for the decision.
Herbert Simon, a theorist of organizations and decision-making, pointed out that decision-making always involves acting with bounded rationality. AI can expand this rationality by providing access to more information, scenarios, and calculations, but it does not eliminate uncertainty, trade-offs, or conflicts of values. It shifts the decision-maker’s workload: less focus on producing all the analyses oneself, and more on understanding what they reveal, what they conceal, and what they entail.
Management will be particularly affected by this transformation. Management will no longer consist solely of organizing people’s work, but of organizing cooperation between people and intelligent systems. Tasks will need to be redistributed, and it will be necessary to determine what can be automated, what must remain a human responsibility, what must be monitored, what must be explainable, and what must be prohibited.
Managers will, in part, become stewards of hybrid intelligences.
This expression may sound futuristic. Yet it describes a reality that is already taking shape. In some teams, AI systems draft initial versions of documents, analyze data, assist with customer relations, prepare decisions, monitor risks, and automate processes. In the future, specialized agents will collaborate to carry out complex task chains. Human work will not automatically disappear, but it will shift in focus. It will increasingly involve design, decision-making, supervision, relationship-building, creativity, explanation, and accountability.
This shift will also have an impact on recent graduates. For a long time, the early stages of a career often involved performing preparatory tasks: conducting research, producing summaries, formatting documents, analyzing simple data, and preparing materials. Yet it is precisely these tasks that AI is beginning to take over. We will therefore need to rethink how people enter the workforce.
If the early stages of vocational training are automated, how do we cultivate experience? How do we learn a trade when the tasks that used to teach it are disappearing or changing?
This is a crucial question. You don’t become an expert simply by being entrusted with lofty responsibilities. You also become one through gradual exposure to the nitty-gritty details, mistakes, imperfect files, ambiguous instructions, difficult clients, messy data, and day-to-day trade-offs. If AI takes over this raw material of learning too early, organizations risk creating professionals who are quick but superficial; productive but inexperienced; supported but fragile as soon as the situation deviates from the expected framework.
We will therefore need to develop new forms of professional learning. Rather than letting young employees use AI to avoid tasks that provide training, we should teach them to use AI to better understand what those tasks entail. A generated summary can become a subject for analysis. Code generated by AI can serve as a tool for understanding. An automated recommendation can provide an opportunity to discuss assumptions, risks, and limitations. Professional mentorship isn’t going away; it must incorporate the machine as a new player in the learning journey.
Donald Schön, a theorist of professional learning, has highlighted the importance of the reflective practitioner—someone who learns by reflecting on their actions and through their actions. This idea becomes essential in the age of AI. The augmented professional will not be the one who clicks faster. It will be the one who knows how to observe their own thinking alongside the machine, understand how it alters their reasoning, and identify the moments when it helps them and those when it lulls them into complacency.
It is also important to highlight a major managerial risk: the illusion of productivity. Generative AI can give the impression that an organization is producing more simply because it is generating more documents, more messages, more tables, more scenarios, and more presentations.
But an organization can produce more content while generating less meaning. It can speed up the flow of responses while slowing down the quality of decisions. It can fill its servers with flawless text while draining its meetings of genuine thought.
Increased productivity must not, therefore, lead to an explosion of paperwork.
The purpose of intellectual work is not to produce artifacts more quickly. Its purpose is to understand better, make better decisions, serve better, create better, and cooperate better.
If AI speeds up tasks without improving their outcomes, it does not transform work; it merely makes it more elegant.
That is why organizations will need to establish a genuine culture of AI governance. This governance cannot be reduced to a charter forgotten on an intranet. It must define authorized uses, risk levels, responsibilities, verification requirements, data protection, explainability rules, and situations where humans must take control. It must also include training, because you cannot sustainably govern a technology that teams do not understand.
Luciano Floridi, a philosopher specializing in the ethics of information, emphasizes the need to view our digital environments as spaces in which our responsibilities, identities, and actions are reshaped. This perspective is valuable for businesses.
AI is not merely a tool added to an organization. It transforms the very nature of work: the flow of knowledge, power dynamics, decision-making processes, forms of oversight, performance expectations, and the boundaries between assistance and replacement.
The issue of work in the age of AI is therefore not merely a question of which jobs will disappear or emerge. While this is an important question, it is not the whole story. The deeper question concerns the quality of the human work that will remain, and the role we want it to play.
Do we want organizations where AI is used to increase surveillance, accelerate processes indiscriminately, standardize behavior, and turn every employee into an algorithm-driven cogs in the machine? Or do we want organizations where AI frees up time to better understand, better support, better create, better decide, and better nurture human relationships?
This choice is not a technical one. It is cultural and political.
In the professional world, AI will not simply eliminate tasks. It will shift the focus to what machines cannot do: judgment, meaning, relationships, and responsibility.
That is why the most sought-after skill may not just be knowing how to use AI, but knowing how to remain human in an increasingly automated environment.
This means being able to question a recommendation, challenge a prediction, delay a decision when its consequences are too significant, explain a choice to those affected by it, and preserve a relationship when the system pushes for standardization.
The professional of tomorrow will not be someone who works against AI, nor someone who works under AI. It will be someone who knows how to work with it without relinquishing their own judgment.
They will no longer be merely producers of deliverables. They will become orchestrators of challenges, knowledge, tools, relationships, and responsibilities. And this transformation will require companies to rethink their hiring practices, training programs, management styles, evaluation criteria, and very concept of performance.
Because a smart organization isn’t one that has simply deployed the most AI tools. It’s one that has used them to elevate the quality of its collective judgment.
IV. Social Risk: The Divide Between Those Who Use AI and Those Who Think With It
Redefining intelligence will not only have an impact in schools and businesses; it will also reshape the social landscape.
It is often believed that the major divide of the future will be between those who have access to AI and those who do not. This divide will certainly exist, and it must be tackled with determination. But it may not be the deepest one. Access to these tools will likely become widespread over time, just as smartphones, search engines, and social media have become widespread. The real divide will be more subtle, more silent, and therefore more dangerous.
It will distinguish between those who use AI as a source of answers and those who use it as a demanding intellectual partner.
The former will ask the machine to do the work for them. The latter will learn to think more effectively with it. The former will save time, sometimes at the expense of their autonomy. The latter will gain intellectual power, because they will know how to question, verify, connect, and decide. The former risk becoming dependent on an external intelligence. The latter will transform that external intelligence into a tool for discernment.
The great divide of the future will not merely separate connected people from disconnected people. It will separate those who think with AI from those who think by proxy.
This is a major issue, because AI can either serve as a powerful catalyst for empowerment or a potent mechanism for disempowerment. Everything will depend on the intellectual, ethical, and social culture that shapes its use.
Amartya Sen, an economist and philosopher of human capabilities, has shown that development is not merely about access to resources, but about individuals’ actual ability to convert those resources into concrete freedoms. This idea is crucial to understanding AI. Having access to a tool does not mean knowing how to use it to think, learn, create, work, or make decisions freely.
Access is a prerequisite. It is not emancipation.
A student who has access to AI but cannot evaluate its responses remains vulnerable. An employee who uses an intelligent assistant without understanding its limitations remains dependent. A citizen who receives generated content without knowing how to question its sources becomes more susceptible to manipulation. An organization that automates without governance may gain speed, but risks losing accountability. A country that uses systems designed elsewhere, trained elsewhere, and aligned with other values may gain apparent efficiency while losing some of its intellectual sovereignty.
The issue, then, is not merely a digital one. It is a democratic one.
In a society saturated with content, manipulation no longer necessarily needs to hide information. It can simply bury it under plausible answers, convincing images, well-crafted narratives, and manufactured certainties.
Democratic fragility no longer stems solely from censorship or a lack of information. It also stems from an excess of indistinguishable content.
Hannah Arendt, a philosopher of politics and responsibility, understood that factual truth is a fragile prerequisite for communal life. When facts become blurred, when everyone can retreat into a narrative that is perfectly coherent but detached from reality, democratic discourse breaks down. Generative AI makes this issue even more pressing, as it has the potential to mass-produce plausibility.
That is why training in AI cannot simply mean learning how to use tools. It must mean learning to remain free in an information environment that has become extraordinarily powerful. Free to doubt. Free to verify. Free to reject an answer. Free to understand the interests that shape these systems. Free not to confuse personalization with truth, efficiency with justice, or automation with progress.
Freedom cannot be decreed. It is something that develops.
It requires teaching children to exercise good judgment from an early age, ongoing training for professionals, capacity-building for leaders, media accountability, high standards for public institutions, corporate governance, and a renewed culture of civic engagement.
Because AI will not be just a topic for specialists. It will become an invisible infrastructure of our everyday lives. It will shape the way we learn, shop, work, stay informed, receive medical care, hire, make decisions, and perhaps even vote in the future. When a technology becomes this pervasive, ignorance is no longer just a setback. It becomes a collective vulnerability.
States, schools, universities, businesses, and families will therefore bear a historic responsibility. Failing to provide widespread training in AI is not merely accepting a technological lag; it is accepting cognitive dependence. It means allowing others to define the tools our children will use to learn, our employees to work, our citizens to stay informed, and our organizations to make decisions.
The sovereignty of the future will not be limited to industry, energy, or the military. It will also be cognitive.
It will depend on our collective ability to understand the systems that guide our choices, to manage the tools that shape our decisions, and to maintain our judgment in a world where responses have become automatic.
That is why the issue of intelligence is not an abstract one. It touches on social justice, democracy, sovereignty, and human dignity. If we fail to bring about this transformation, AI risks creating a two-tiered humanity: on one side, those who will be able to enhance their thinking, and on the other, those who will gradually delegate their judgment.
In this new hierarchy, the danger will not simply be that of being replaced by a machine. It will be more subtle, more profound, more personal: losing control over one’s own thoughts, choices, and future.
Conclusion: Toward a New Humanism of Intelligence
We must therefore move beyond this false debate.
The issue is not whether artificial intelligence will replace human intelligence. Framed this way, the question traps us in a narrow-minded rivalry. It assumes that humans and machines are playing the same game, on the same playing field, with the same goals.
But that is not the case.
AI processes information, but it does not inhabit the world. It calculates possibilities, but it has no lived memory. It generates responses, but it is not morally accountable for what it produces. It can simulate a conversation, but it does not share our vulnerability. It can generate text about love, fear, death, justice, or hope, but it does not experience any of these things.
That is why the humanism we need cannot be a humanism rooted in nostalgia. It is not a matter of defending humanity as a citadel under siege by machines. Nor is it a matter of repeating that humans are superior in every way, as if that alone were enough to reassure us.
That would be little consolation.
The new humanism will have to be more demanding. It will have to clearly recognize what machines already do better than we do, without concluding that we are worth less than they are. It will have to accept that certain human capabilities are being surpassed, without confusing capability with dignity. It will have to teach us to cooperate with AI without losing ourselves in it.
Martha Nussbaum, a philosopher of humanistic education and capabilities, has often emphasized that education should not merely cultivate skills useful to the job market, but should also foster individuals capable of judgment, moral imagination, and democratic participation. This idea becomes central in the age of AI. The more powerful machines become, the more we must cultivate what they cannot provide in our place: conscience, responsibility, and sensitivity to the shared world.
The danger, then, is not that machines will become intelligent. The danger is that we will accept a narrow definition of our own intelligence. If we reduce humanity to its capacity to produce, then AI will humiliate us. If we reduce intelligence to speed, then AI will surpass us. If we reduce work to mere execution, then AI will replace us. If we reduce education to rote memorization, then AI will render it obsolete.
But if we redefine intelligence as the ability to question, discern, make sense of things, connect ideas, and take responsibility, then AI will not merely be a threat. It will become a litmus test. It will force us to raise our standards. It will compel us to become less mechanical in the way we think, learn, work, and make decisions.
This may be the most fruitful paradox of this revolution: machines are forcing us to become more human again.
Not humans against them, but humans alongside them. Not humans diminished by their power, but humans reminded of what defines our uniqueness. Not humans protected by ignorance, but humans made more responsible by understanding.
Maybe my son was right that night. Maybe ChatGPT explained it to him more clearly than I did. There’s no point in denying it, and it would probably be rather futile to try to reclaim my authority as a father by trying to rephrase things on the spot.
But that's not where the real issue lies.
The question is no longer whether a machine can explain a lesson, write a summary, correct an exercise, or produce a convincing answer. The question is whether we will still be able to pass on to future generations what a machine cannot decide on its own: the quality of questions, the prudence of judgment, the depth of meaning, the art of connecting worlds, and the courage to stand by one’s choices.
We don’t need to train people to imitate machines. We need to train people who can guide them.
Because, ultimately, AI does not merely raise questions about the future of technology. It poses a question that is older, more delicate, and more decisive: what do we still want to call intelligence, so that our children do not merely become high achievers, but fully human beings?
References
- Arendt, H. (1958). The Human Condition. Chicago: University of Chicago Press.
- Bender, E. M., Gebru, T., McMillan-Major, A., & Shmitchell, S. (2021). “On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?” Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency, 610–623.
- Dewey, J. (1938). Logic: The Theory of Inquiry. New York: Henry Holt and Company.
- Drucker, P. F. (1959). Landmarks of Tomorrow. New York: Harper & Brothers.
- Floridi, L. (2013). The Ethics of Information. Oxford: Oxford University Press.
- Kahneman, D., & Tversky, A. (1979). “Prospect Theory: An Analysis of Decision under Risk”. Econometrica, 47(2), 263–291.
- Meirieu, P. (1991). The Choice to Educate: Ethics and Pedagogy. Paris: ESF.
- Morin, E. (1990). An Introduction to Complex Thought. Paris: ESF.
- Nussbaum, M. C. (2010). Not for Profit: Why Democracy Needs the Humanities. Princeton: Princeton University Press.
- Schön, D. A. (1983). The Reflective Practitioner: How Professionals Think in Action. New York: Basic Books.
- Sen, A. (1999). Development as Freedom. New York: Alfred A. Knopf.
- Simon, H. A. (1957). Models of Man: Social and Rational. New York: Wiley.
- Turing, A. M. (1950). “Computing Machinery and Intelligence”. Mind, 59(236), 433–460.
- Vygotsky, L. S. (1978). Mind in Society: The Development of Higher Psychological Processes. Cambridge, MA: Harvard University Press.
Assessing Your Readiness for the New Human Intelligence
This manifesto puts forward a conviction: AI does not merely require us to use new tools; it compels us to redefine what we call intelligence. But a conviction alone is not enough. We must be able to translate it into practices, criteria, decisions, and action plans.
That is why I have developed a set of practical tools that enable higher education institutions, businesses, and individuals to assess their level of maturity across the five new dimensions of human intelligence in the age of AI: questioning, discerning, making sense, connecting, and taking responsibility.
These tools do not simply measure the level of technological infrastructure or the number of AI applications deployed. They measure something far more essential: the actual ability of an organization or an individual to think alongside AI without relinquishing their own judgment.
Does a higher education institution truly train its students to use AI methodically, with critical perspective, and responsibly? Do its assessments still measure only the final answer, or do they also evaluate the intellectual process that leads to it? Are faculty members supported in integrating AI into their teaching practices without compromising academic standards? Do the curricula foster the ability to ask questions, exercise discernment, make sense of information, establish connections, and act responsibly?
A company: Does it use AI solely to speed up production, or to improve the quality of its decisions? Do its employees know how to verify, contextualize, and take ownership of the outputs generated? Do managers know how to organize work between humans and intelligent systems? Do hiring and training plans take into account the new critical skills required in the age of AI?
An individual: Do they use AI merely as a source of answers, or as a true thinking partner? Do they know how to ask better questions, identify the limitations of an answer, connect different perspectives, protect sensitive data, and retain ownership of what they produce?
These assessment frameworks help establish a clear diagnosis, identify areas of weakness, determine an organization’s level of maturity, and develop a step-by-step roadmap. They can be used as a self-assessment tool, a training resource, a basis for strategic planning, or a starting point for more structured support.
This leads to a simple classification: ranging from the spectator-oriented organization to the enhanced humanistic organization, or from the naive user to the responsible creator. The goal is not to classify in order to punish, but to highlight areas that need improvement.
Organizations or companies that wish to access the organizational diagnostic frameworks, adapt them to their specific context, or discuss their implementation can request a callback using the following form: https://aivancity.ai/demandez-etre-rappele .
Anyone can also test their own level of AI literacy for free using the individual assessment available here: https://aivancity.ai/grille_indiv_maturite_IA/ . In just a few minutes, it provides a personalized assessment of your ability to question, discern, make sense of, connect, and take responsibility for your use of AI.