Craig Barton interviews guests from the wonderful world of education about their approaches to teaching, educational research and more. All show notes, resources and videos here: https://www.mrbartonmaths.com/blog/
A decade after his first appearance on the podcast, Dan Meyer returns to discuss the intersection of AI and mathematics education. Now leading AI feature development at Amplify (which acquired Desmos), Dan brings a uniquely balanced perspective—simultaneously a critic of AI's most maximalist claims and an active builder of AI tools for the classroom. Craig and Dan dig into three core areas: AI as a personal tutor, AI for teacher professional development, and AI for assessment. Along the way, they explore why AI tutors keep falling short of their hype, what Amplify's ”discussion moments” feature reveals about thoughtful AI integration, the realities of the Alpha School model, and the social dimensions of learning that no chatbot can replicate. Dan argues that genuine educational improvement may depend less on technology and more on political solutions like teacher pay, class sizes, and addressing inequality—a sobering counterpoint to the dominant AI-as-saviour narrative. the show notes here: podcast.mrbartonmaths.com/220-ai-in-education-with-dan-meyer/
Episode sponsor
Maths teachers have built a free ChatGPT app that can generate worksheets directly in chatGPT, with proper mathematical notation, graphs and even diagrams!
To get started, search “Worksheet” in the ChatGPT App store. Prompt the Strive PDF Generator to create whatever you need: unit tests, homework or quizzes, then just download the fully-formatted PDF. No latex knowledge or copy-paste required.
Go to worksheet.chat to get started.
Episode details
A decade after his first appearance on the podcast, Dan Meyer returns to discuss the intersection of AI and mathematics education. Now leading AI feature development at Amplify (which acquired Desmos), Dan brings a uniquely balanced perspective—simultaneously a critic of AI’s most maximalist claims and an active builder of AI tools for the classroom. Craig and Dan dig into three core areas: AI as a personal tutor, AI for teacher professional development, and AI for assessment. Along the way, they explore why AI tutors keep falling short of their hype, what Amplify’s “discussion moments” feature reveals about thoughtful AI integration, the realities of the Alpha School model, and the social dimensions of learning that no chatbot can replicate. Dan argues that genuine educational improvement may depend less on technology and more on political solutions like teacher pay, class sizes, and addressing inequality—a sobering counterpoint to the dominant AI-as-saviour narrative.
Talking points
- Dan’s journey from secondary maths teacher to Stanford doctorate to leading AI development at Amplify, and how classroom visits keep him grounded against “AI psychosis”
- Why Dan uses AI enthusiastically for coding and prototyping but never for his writing—and the John Warner argument that “writing is how we think”
- The collapse of the AI tutor revolution: even Sal Khan has admitted Khanmigo isn’t delivering what was promised
- The hard problem of intervention timing: how skilled teachers decide when to step in versus let students struggle, and why this seems unsolvable for chatbots
- Craig’s pushback using Eedi’s RCT findings: hallucinations and latency may be more tractable than feared, and some students prefer asking AI over teachers due to embarrassment
- The “95% problem”—why personalised learning efficacy studies often only show gains after excluding the vast majority of students
- A deep dive into Amplify’s “discussion moments” feature: how AI surfaces interesting student responses to support whole-class discussion rather than replacing the teacher
- The submarine integers lesson as a case study in low-floor pedagogy and eliciting mathematical creativity
- Dan’s two-part theory of teaching: invitation and development—and how this bridges progressive and direct instruction camps
- Why students show off for peers and teachers but not for LLMs, and what that means for cognitive engagement
- An honest assessment of Alpha School: interesting questions being asked, but top-of-market tuition pre-selects for parental motivation in ways that limit generalisability
- US vs UK classroom comparisons: teacher variability, curriculum prescription, and cultural differences in maths instruction
- Where AI genuinely helps teachers: generating practice problems, isomorphic test questions, and headache-aspirin scenarios
- The “10x for experts, 0.9x for novices” problem in AI-assisted teaching
- Handwriting analysis as a promising frontier where AI doesn’t threaten the social fabric of the classroom
- Dan’s provocative closing argument: why we keep reaching for AI solutions when increasing teacher pay and reducing class sizes have stronger evidence bases
- A sobering vision of the future shaped less by technology than by widening inequality and political choices about how we resource schools
Video:
Links from Dan
- Dan’s Substack, Math Worlds, is here: danmeyer.substack.com
- You can check out Dan’s company, Amplify, here: amplify.com
- And the classic Desmos graphing tool here: desmos.com/calculator
New stuff I have been working on:
- My Tips for Teachers Guides to… series
- My updated mrbartonmnaths website
My usual plugs
- You can help support the podcast (and get an interactive transcript of this episode) via my Patreon page at patreon.com/mrbartonmaths
- If you are interested in sponsoring an episode of the show, then please visit this page
- You can sign up for my free Tips for Teachers newsletter and my free Eedi newsletter
- My online courses are here: craigbarton.podia.com
- My books are “Tips for Teachers“, “Reflect, Expect, Check, Explain”, “How I wish I’d taught maths”, and my Tips for Teachers Guide to… series.