AI in Schools 2026: Predictions, Policy, and What Might Actually Scale

For show notes, links, and a summary episode, sign up for the ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Hey! What You Reading For ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠newsletter. Mondays at 7am BST - https://tdape.beehiiv.com/subscribe

For maths curriculum questions contact us ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠here⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠ or via support@alta-education.com

Learn more about ⁠⁠⁠⁠The Story of Maths⁠⁠⁠⁠ - www.alta-education.com/tsom-overview

Episode 262: In this episode of Thinking Deeply about AI for Schools, James Radburn and Neil Almond start by looking back at “Education Technology Trends to Watch in 2025” and ask what truly moved, what stayed theoretical, and why AI-driven personalised learning still hasn’t landed at scale. They dig into a crucial question: what do we even mean by “personalised”—better sequencing and timing of knowledge, or just swapping examples based on interests?

Next, they run a fun (and revealing) experiment: asking multiple LLMs—ChatGPT, Grok, Claude, Gemini, DeepSeek—what they predict about AI in schools, from AI detection dying off to AI showing up more in policy than classroom practice.

Then the hosts give their own six predictions for 2026, covering lots of possibilities for the next 12 months.

A practical, slightly sceptical, educator-first episode about what’s next—and how schools can innovate without getting governed by AI instead.