The Point of Prediction Is to Prepare

Humans have a long record of predicting the future incorrectly. The problem is not a lack of intelligence or imagination; too many variables interact, and small changes compound in ways no one can foresee.
Still, predictions matter. Their value is not that they make us infallible about the future. It is that they force us to prepare for more than one future.
I spend a fair amount of time thinking about how AI will affect education and the transition from learning to work. I have to. Education, skills, and pathways to employment sit at the center of IREX’s work. They are also where many of AI’s consequences for people will become tangible. In that spirit, here are a few predictions, loosely held.
First, the AI revolution may not unfold with a bang. It may look more like an evolution. The next phase is one in which AI becomes a routine layer of learning and work, embedded in classrooms, universities, workplaces, and the systems people already use. That transition is already underway.
Second, more AI will make human judgment more valuable, not less. When machines can generate competent answers, summaries, code, and analysis in seconds, production becomes abundant. Judgment becomes scarce.
That can sound abstract, but we see its importance in our leadership work. Information and technical knowledge matter, but leadership ultimately requires judgment: weighing competing interests, acting without complete information, taking responsibility, and deciding what is right when the answer is not obvious. AI can inform those choices. It cannot own them.
Third, institutions will increasingly be seen as hurdles. Students, teachers, and workers are already using AI, and many will find that the schools, universities, employers, and governments around them move more slowly. That does not necessarily mean they are resistant to AI. Institutions operate within the same constraints that shape their effectiveness more broadly: bureaucracy, competing interests, political economy, incentives, uneven capacity, and, in some contexts, corruption. These forces can slow adaptation even when the case for change is clear. Schools, universities, employers, and governments will become more intentional about their use of AI, but many will struggle to put in place the governance and leadership needed to harness AI for their own purposes. The challenge is not simply adopting AI. It is knowing where AI adds value, what capabilities are needed to use it well, and where human judgment must remain in control.
Finally, economics will matter, as it already does. My point is twofold. First, today’s investment in AI rests on expectations of massive future returns. The bursting of the dot-com bubble offers a useful cautionary note: a new technology can transform the world even as the investment surrounding it gets ahead of economic reality. Second, whether or not there is a correction, AI will increasingly arrive on unequal terms. Language coverage, connectivity, teacher preparation, data governance, and purchasing power will shape who benefits. A student learning in a widely supported language, with reliable connectivity and an AI-ready teacher, starts from a different place than one without those advantages. New technologies can expand opportunity and widen divides at the same time. AI will be no different. The lesson and opportunity are to act early, while there is still time to shape how widely those benefits are shared.
I do not know which of these predictions will prove right. Our job is not to predict one future correctly. It is to be ready for several, and that is the point of prediction.
Check out recent editions of Global Outlook.
A Different Kind of Leadership for a Polarized Age draw from programs and discussions convened on the margins of the United Nations General Assembly (UNGA) and are offered for their timely relevance.
From Jordan to the World: Practical Lessons for Early Grade Reformers explores practical lessons from IREX and its partners on strengthening foundational learning through coherent, classroom-focused reform.