Close the Gap or Widen It? AI, Equity and Learning at the Weakest Link

Generative AI could be the most powerful equaliser in the history of education, putting a patient, personalised tutor within reach of every child. It could equally widen the gap between advantaged and disadvant‐ aged learners, because new technologies usually reach those with re‐ sources first and those without last, if at all. This article draws on a practitioner newsletter series written for school leaders in 2026, read alongside data on connectivity, refugee education, aid and AI develop‐ ment policy. It argues that the outcome will not be decided by the techno‐ logy but by deliberate choices, and that leaving AI to markets and chance will produce the worse outcome. It identifies four layers of the AI divide: access, skilled guidance, language and culture, and recognition of what learners achieve. It then examines the weakest link in global education, learners in crisis and displacement, where 5.8 million refugee children were out of school in 2025 and international aid fell by nearly a quarter in a single year. It reviews approaches that work where power, connectivity and funding cannot be assumed, including offline platforms, AI delivered through basic phones, support for community teachers and portable, re‐ cognised certificates, along with their risks. It proposes a six-question equity test for any AI decision in education.