Education World Forum 2026: how do we educate children for jobs that don't yet exist?

As a Master in Public Policy (MPP) student at the Blavatnik School of Government, University of Oxford, I visited the Education World Forum alongside the What Works Hub for Global Education. As the world's largest annual gathering of education ministers, bringing together over 150 ministers registered and more than 1,500 delegates representing 125 nations, it is an opportunity for the What Works Hub for Global Education to engage in dialogue with decision makers on the implementation of effective education policy at scale.

Themed Educating for a Shared Future: Peace, Planet, Purpose and Pathways, this year's forum explored how education systems can be reimagined to address global challenges while supporting long-term economic growth, inclusion and sustainability. Every policy professional needs a deep understanding of pressing issues, and attending gave me much greater clarity on the challenges and opportunities shaping global education today. Being in a room where governments compare reforms, discuss shared challenges and build partnerships around education systems was, for me, an invaluable extension of the MPP experience.

The AI conversation

One strand that came through consistently across the sessions was the urgent need for education that is ahead of its time, equipping children not merely for the world as it is, but for the world as it will be. Artificial intelligence sat at the heart of this conversation. At a plenary session, George Osborne, MD of OpenAI for Countries, noted the rapid spread of ChatGPT, which has been publicly reported to have around 900 million weekly active users. OpenAI has also identified “learning and upskilling” as a major use case, accounting for 20% of usage in one US-focused analysis. This figure underscores how rapidly AI has embedded itself into education systems, regardless of whether those systems are ready for it. Beyond student use, AI is also being explored as a tool for teacher coaching, a means of training teachers in less time and at greater scale. The consensus, however, was that doing this well matters enormously. Deploying AI thoughtfully in teacher development can meaningfully strengthen instructional quality; poorly designed interventions risk reinforcing existing gaps rather than closing them.

The inequality problem

The conversation around AI in education, however, sits uneasily alongside a reality I encounter in my own work with children in rural communities under the umbrella of an  NGO I cofounded - Grassroots Initiative for Positive Socioeconomic Development (GIPSED). These children, already navigating deep educational disadvantages, are among those least likely to benefit from AI-driven learning tools and most likely to be further left behind by them. Duncan Crawford, Senior Content Manager at the OECD, presented a finding at the forum that left a strong impression on me: by age five, inequalities in learning and development are already measurably widening. Before most children have even started primary school, the gap is already opening.

Equally striking is how layered the barriers really are. It is not simply about who has a device and who does not. It runs deeper, from language, where AI tools are predominantly built for a narrow set of languages that exclude millions of learners, to history, where entire regions of the world lack the digitised cultural and knowledge base that makes these tools meaningful and relevant. Access, in other words, is not just a question of hardware. It is a question of whether the technology was ever designed with these learners in mind.

A warning for the global south

This is where my concern deepens considerably. There is a growing pressure on countries in the global south to adopt AI frameworks that were designed elsewhere, by others, for different contexts. But deploying sophisticated tools onto weak or under-resourced education systems does not accelerate learning, it accelerates the appearance of progress while the underlying problems remain untouched. AI must be an amplifier of strong foundations, not a cosmetic layer over crumbling ones. My conviction, leaving the forum, is that countries in the global south must resist this pressure and find the confidence to define their own visions for what AI should do in their education systems, on their own terms, at their own pace, with their own priorities.

The human element

Perhaps the highlight of the discussion at the Forum that resonated most deeply with me, however, was the one about what we must not lose in all of this. As AI takes on more functions in the classroom with personalised feedback, progress tracking and predictive analytics, there is a quiet danger that the human dimensions of education are slowly being designed out. While I do not think this is anyone's intention, the risk is real and growing. The relationships between teachers and students, the friction of collaborative problem-solving, the experience of learning to navigate disagreement with peers; these are not soft extras sitting alongside the real work of education. They are the real work. It is precisely in these moments of human exchange that the most durable learning takes place, the kind no algorithm can replicate. In a world where technical tasks are increasingly automated, these human capacities may ultimately prove to be the last truly irreplaceable competency.

How we navigate that balance will define the quality and equity of education for the next generation. The stakes could not be higher as we are no longer preparing young people for known occupations, we are equipping them for roles that do not yet exist, in a world that refuses to hold still. That is a challenge that, judging by the conversations at EWF 2026, no country is yet to fully resolve.