How might low- and middle-income countries plan intentionally for AI to contribute to long-term improvements in student learning rather than short-term acceleration?
EdTech Hub’s AI Observatory is exploring how education systems can renew the purpose of learning — which means rethinking what education is for in a world shaped by AI, with the aim of narrowing the learning divide.
This week, in Issue No. 32 of the #WaypointWednesday, we spotlight ways that education systems are using AI to support the development of skills, designing for pedagogy first, and considering AI-free learning environments.
Early signals
AI-integrated assessments
We’re seeing assessments that incorporate AI in different ways — from large-scale diagnostic assessments that inform instruction, to oral examinations that help verify student understanding.
- India – Competency-based diagnostic assessments:Rajasthan’s education department is moving away from traditional exams toward AI-enabled, competency-based diagnostic assessments aimed at building enduring skills. AI-powered tools were used to assess over 4.5 million students in Hindi, English, and Maths, with the insights driving 270 minutes of structured weekly remediation to support sustained learning progress. (Digital Learning Elets, June 2025)
- Global – AI-supported oral assessment: The AutoViva system uses generative AI (GenAI) to support oral examinations, where students present their work and respond to questions from examiners. This approach draws on the long-established viva voce method of verifying complex coursework, while attempting to make it scalable. Early evaluations suggest high student acceptance and the ability to manage large numbers of participants simultaneously. (Cao and Zahid, 2025)
Pedagogy-first design
We’re seeing a reframing of AI in education so that pedagogy — not automation — sets the terms of how AI is designed and used in learning environments. (UNESCO, 2025)
- Global – OECD Digital Education Outlook 2026:OECD reports that while general-purpose GenAI tools can improve students’ performance on specific tasks, this does not necessarily translate into learning gains. In contrast, AI tools designed or used with clear pedagogical intent are more likely to support sustained improvements in learning outcomes. (OECD, 2026)
- Qatar – Feature fallacy: Novel AI features do not always translate into meaningful learning. For instance, forthcoming research from the WISE AI Testbed Cycle 1 Evaluation Report found that converting dense reading materials into AI-generated audio podcasts led to passive consumption and disengagement among students. (Moustafa et al., forthcoming)
AI-free learning environments
Ben Williamson observes “a fast-growing body of writing” arguing that AI may be incompatible with education. One response emerging from this debate is the call for learning environments where AI is deliberately limited or excluded. (Williamson, 2026)
- UAE – Age restrictions on GenAI: Education authorities in the UAE have introduced policies restricting the use of generative AI tools for students under the age of 13 or those enrolled below Year 7. The decision aims to protect early-stage learning processes based on interaction, creativity, and independent skill development. (Gulf News, 2026)
- Global – The case for AI illiteracy: O’Sullivan argues that what is often framed as “AI literacy” may unintentionally obscure forms of knowledge that academic practice seeks to preserve. He suggests that strategic refusal or “productive ignorance” of AI systems may sometimes be intellectually valuable, particularly in disciplines centred on interpretation, critique, and originality. (O’Sullivan, 2025)
Reflections
- To support students in developing long-term skills, AI that privileges the shortest learning path, may not be the most beneficial in the long run. Deep learning requires students to struggle with concepts, tackle a challenge in different ways, obtain multiple perspectives, and grapple with difficulty. (Adam & Lester, 2026)
- AI-free learning spaces are critical to help students reconnect with what it means to be a human and an independent critical, and creative thinker. Similar to digital detoxes that can improve clarity in thinking and dopamine regulation, AI-free spaces can support restoring cognitive efforts, slower and deliberate thinking, and rebuild comfort with uncertainty. (Adam, 2026)
