Futures in Context
Mike Garner
Head of Creative Strategy, Connelly Partners | Writer | Creative Strategist | Creative Director
Mike Garner is a strategist and writer with forty years of experience at the intersection of culture, communication, and ideas – having built creative organisations for Saatchi & Saatchi, OgilvyOne, and his own agency, Chemistry. His WPP Atticus Award-winning essay Here Be Dragons established him as a distinctive voice on creativity and its limits. He works closely with Liam Wegimont and the Global Education Network of Europe on critical thinking, global citizenship, and the human capacities that an AI-saturated world most urgently needs.
AI & EDUCATION — CHAPTER 3
In November 2022, two significant things happened within days of each other.
In Dublin, representatives from across Europe adopted the Dublin Declaration on Global Education to 2050—a commitment to education that opens “eyes, hearts, and minds to the reality of the world.”
In San Francisco, OpenAI released ChatGPT to the public. Both were responses to the same underlying pressures: a world of escalating complexity, contested truth, and fraying social cohesion. But they embodied radically different ideas about how humans might navigate what lies ahead.
Two years on, the relationship between these two developments matters more than most people realize. AI has transformed the information landscape in ways that make Global Education more essential than ever. And more threatened.
The double bind
Large language models have collapsed the cost of producing plausible content to near zero. This isn’t simply a “fake news” problem in the narrow sense. The deeper issue is the flooding of information environments with material that may be technically accurate but is strategically misleading, emotionally manipulative, or designed to incite rather than inform.
And crucially, AI translation now enables misleading narratives crafted in one language to reach global audiences within hours, adapted to local contexts with sophisticated cultural tuning. Misinformation has gone multinational.
The technology industry’s response has been to offer more AI. Fact-checking algorithms. Content moderation systems. Accuracy labels generated by machine learning. These tools treat misinformation as a content problem: identify the false claims, label or remove them, problem solved. But this misses the point entirely.
Misinformation thrives not because people lack access to accurate information. It thrives because they lack the capacities and dispositions to engage with complexity critically. Algorithmic fact-checking addresses symptoms while leaving the underlying vulnerability untouched. And here’s the part nobody wants to say out loud: Every time we outsource evaluation to a machine, we weaken our own judgment. The muscle atrophies through disuse.
There’s a distinction worth being precise about here. The gap between understanding and comprehension. An AI fact-checker can tell you a claim is false. It cannot help you understand why you might be susceptible to it, what emotional or social needs it addresses, or how to engage constructively with someone who actually believes it.
It facilitates understanding—accurate transmission of information—but not deep comprehension: the integration of knowledge with values, context, and the hard-won capacity for judgment. That gap is precisely where Global Education lives.
Misinformation thrives not because people lack access to accurate information. It thrives because they lack the capacities and dispositions to engage with complexity critically
Not all information is the same kind of thing. There are at least four qualitatively different levels and the distinction matters enormously. Syntactic information is context-free: a bit, a data point, a yes or no. Semantic information carries meaning, but meaning requires a frame of reference. Pragmatic information requires a situated agent—a person, in a body, making a decision under conditions of uncertainty. And phenomenal information is conscious, embodied, first-person experience itself: the felt quality of what it is actually like to understand something from the inside.
AI operates with extraordinary power at the first two levels. It can process and transmit data and meaning at scale that dwarfs anything a human can manage. But it has no access to levels three and four. It cannot be situated. It cannot feel. Global Education operates precisely where AI cannot: at the levels of pragmatic judgment and lived comprehension that are the foundation of genuine critical thinking. This is not a limitation that better technology will eventually overcome.
It is a structural boundary between what machines can process and what humans must develop for themselves.
What global education knows
The Dublin Declaration speaks of education that enables learners “to reflect critically on the world and their place in it” and empowers them “to understand, imagine, hope, and act.”
Note the structure. Understanding is necessary but not sufficient. It must connect to imagination, the capacity to envision alternatives. To hope, the emotional foundation for sustained engagement. And to action, the translation of comprehension into citizenship.
AI can support understanding. It struggles with imagination. It cannot generate hope. And it can actively undermine the development of judgment when over-relied upon. On the other hand, Global Education practitioners have been developing critical-thinking capacities in areas of contested knowledge, such as climate change, migration, colonial history, and economic inequality, for decades. These are precisely the domains where misinformation thrives, and where algorithmic fact-checking is most inadequate.
What practitioners have learned is that critical thinking isn’t a generic skill you can teach in isolation. It develops through engagement with real controversies, exposure to genuinely different perspectives, and supported practice in holding complexity without retreating into false certainty or paralyzing relativism.
The relational dimension matters enormously. People rarely change their minds because an algorithm tells them they’re wrong. They change through relationships, through dialogue, through the experience of being genuinely heard before being challenged. Global Education’s methodology—participatory, dialogical, values-explicit—is designed for exactly this. No chatbot replicates it.
Think about the capacities that no AI can develop in a learner.
- Tolerance of ambiguity – the ability to sit with questions that don’t resolve cleanly; built through repeated exposure to genuine complexity, not through receiving better summaries of it.
- Epistemic humility – knowing the limits of your own knowledge and perspective, emerges from real encounters with minds that work differently from your own.
- Perspective-taking – genuinely inhabiting a worldview not your own; requires the kind of sustained human dialogue that no algorithm can replicate. Emotional regulation in the face of disturbing information.
- Solidarity – the felt sense of connection with distant others that actually motivates sustained action over time.
These are not information deficits. No AI can fill them. They are human capacities requiring human development, through human relationships, over time. The philosopher Michael Polanyi put his finger on something essential when he observed that “we know more than we can tell.”
The most important forms of human knowledge are tacit: they can be exercised but never fully articulated. The diagnostician’s clinical eye. The experienced teacher’s reading of a classroom. The citizen’s sense that something is wrong before she can say exactly what. These are forms of knowing that are built from first-person experience. They cannot be downloaded. They cannot be transmitted as information. They must be grown through practice, through relationships, through sustained engagement with the messy complexity of the world.
Understanding is necessary but not sufficient. It must connect to imagination, the capacity to envision alternatives. To hope, the emotional foundation for sustained engagement. And to action, the translation of comprehension into citizenship
That is what Global Education provides. And that is what Polanyi’s insight makes clear: No algorithm, however sophisticated, can replicate it.
The convergence moment
We are at a hinge point. The decisions policymakers make in the next few years will determine whether AI serves transformative education or quietly supplants it. The debate is usually framed as a binary: Embrace AI or resist it. Technophile versus technophobe. This framing obscures the real question, which is: What should be the division of labour between AI tools and human educators in developing critical citizens? Here’s a working answer. AI can handle information delivery, translation, accessibility, and content personalization. These are genuine contributions and we should use them.
But humans must remain central to dialogue facilitation, value formation, perspective-taking, and the modeling of engaged citizenship. The danger lies not in AI itself, but in allowing cost and scalability pressures to blur that boundary.
The technology writer Jaron Lanier has articulated the deeper risk with precision: The danger is not that artificial intelligence will become human, but that humans will reshape themselves to become more like AI. Every time a learner outsources evaluation to a chatbot, consults an algorithm instead of forming a judgment, or accepts a generated summary instead of doing the difficult work of reading and thinking for themselves, they are practising not-thinking.
The capacity for critical thought is not a fixed endowment. It is a muscle. It develops through use and atrophies through disuse. An education system that systematically outsources its most demanding cognitive work to machines is not being efficient. It is training its students to be less than they are capable of being.
The real threat to Global Education is not that AI will replace human educators. It is that we will voluntarily abandon the human capacities that Global Education exists to develop—and discover, a generation from now, that we have produced a population fluent in information but incapable of comprehension.
Without intentional policy intervention, education systems will drift toward AI solutions because they are cheaper, more scalable, and more measurable. This pressure is strongest in under-resourced contexts, precisely where Global Education is most needed and least established. The implications are stark. Well-resourced schools will use AI as a supplement while maintaining human-rich education. Under-resourced schools will use it as a replacement. The critical thinking gap will widen along exactly the lines of inequality that already exist, within nations and between them.
Policy implications
Three imperatives follow from this.
- First, protect the irreplaceable. Curriculum frameworks need to explicitly identify the capacities requiring human-mediated development: dialogue, perspective-taking, value formation, engagement with complexity, and guarantee the time and conditions for them. What is easiest to measure is not the same as what matters most. Assessment systems need to reflect that distinction, or they will quietly enforce the wrong priorities.
- Second, defend educational sovereignty. International cooperation on AI governance must include provisions for communities’ rights to determine how their children learn to think critically about global issues. Global Citizenship Education programs must be protected from replacement by AI-mediated content delivery. And Global South educators must be properly resourced to shape AI educational tools, not merely receive them.
- Third, reposition Global Education as AI’s essential complement. Not as its victim. The field has decades of practice, policy, and community to draw on. The task now is to make that case clearly and loudly: that Global Education offers what no algorithm can—not just knowledge about the world, but the formation of people with the wisdom to act within it.
The comprehension imperative
AI can help us understand what others are saying. Translate their words. Summarize their arguments. Check their claims against databases of verified facts. These are useful things.
What AI cannot do is help us comprehend them. Grasp their worldview from the inside. Feel the weight of their concerns. Find within ourselves the capacity for solidarity across profound disagreement.
The danger is not that artificial intelligence will become human, but that humans will reshape themselves to become more like AI
In a world of escalating misinformation and political polarization, that comprehension capacity isn’t a luxury. It is the foundation of democratic society, international cooperation, and collective action on shared challenges.
The Dublin Declaration’s vision—education that opens eyes, hearts, and minds—is not a piece of idealism. It is a description of what is urgently, practically needed. Eyes alone are not enough. Information alone is not enough. Understanding, unaccompanied by the deeper work of comprehension, leaves us fluent but fundamentally unequipped.
The question for policymakers is stark. Will we invest in developing these human capacities? Or will we outsource our thinking to machines and discover, too late, that the ability to process information was never the same thing as the wisdom to act on it?
AI has made Global Education more necessary than ever. The task now is to ensure it doesn’t make Global Education impossible.
