EdTechnical

In this episode of EdTechnical, our podcast hosts Libby and Owen welcome back John Roberts, CEO and co-founder of Oak National Academy, to discuss the results of a new randomised controlled trial of Aila, Oak’s AI lesson assistant. Funded by the Education Endowment Foundation and independently evaluated by the National Foundation for Educational Research, the trial followed 464 primary teachers across 108 schools in England through a term of everyday lesson planning. It asked a simple question: can Aila save teachers time without any loss in the quality of the resources they produce?
Teachers using Aila saved an average of 49 minutes a week, with no difference in the quality of their resources. The entire saving came from just 7% of lessons. This points to a narrow but valuable use case of Aila in planning lessons from scratch, a task that is especially time consuming for early career teachers. John reflects on what this means for Oak’s product strategy and why rigorous evidence struggles to keep pace with the technology it measures. What’s the right balance between saving teachers time and supporting their professional development?
Bio
John Roberts is co-founder of Oak National Academy, the UK government-backed provider of free, open licence teaching resources. He previously led the product and engineering team behind Aila, the UK’s first publicly funded generative AI tool for public use, built to support teachers with lesson planning. Before Oak, John worked as a teacher and senior leader in the North West of England and set up companies and charities, including Teach Your Monster. He is now Oak’s Interim Chief Executive.
Links
Project page and evaluation report by the Education Endowment Foundation: Lesson planning using AI lesson assistant, Aila – Teacher Choices trial
Oak National Academy’s blog on the trial results
Try Aila

Join us on social media: 
BOLD (@BOLD_insights), Libby Hills (@Libbylhhills) and Owen Henkel (@owen_henkel)
Listen to all episodes of EdTechnical here: https://boldscience.org/podcasts/edtechnical/
Subscribe to BOLD’s newsletter: https://bold.expert/newsletter
Stay up to date with all the latest research on child development and learning: https://bold.expert
Credits: Sarah Myles for production support; Josie Hills for graphic design; Pia Schmoeckel for content production.

In this EdTechnical short, Owen is joined by guest co-host Alyssa Van Camp, research lead at EdTechnical, to discuss a recent paper on the learning penalty from generative AI. Drawing on survey data from around 27,000 secondary school students in China, the study looks at how using generative AI may affect homework productivity and learning.
The headline finding is that students who use AI score higher on their homework, but they spend less time on it and their exam results drop. The findings fit with research on how effort shapes learning, though the study design has some limitations. Students self-reported their AI use, time spent on homework was measured indirectly through the school's learning platform, and the paper gives little insight into the quality of the exams used to assess learning.
Surprisingly, around one in five students used AI without a drop in their exam scores. These students spent as much time on their homework as their peers, suggesting they used AI as a tool to help, rather than offload to. What does this mean for designing AI tools that support learning? And how might homework need to change?
Links
Research paper discussed in this episode: The Generative AI Learning Penalty: Evidence from Chinese Secondary Education

Join us on social media: 
BOLD (@BOLD_insights), Libby Hills (@Libbylhhills) and Owen Henkel (@owen_henkel)
Listen to all episodes of EdTechnical here: https://boldscience.org/podcasts/edtechnical/
Subscribe to BOLD’s newsletter: https://bold.expert/newsletter
Stay up to date with all the latest research on child development and learning: https://bold.expert
Credits: Sarah Myles for production support; Josie Hills for graphic design; Pia Schmoeckel for content production.

In this episode, Libby and Owen speak with Joanna Cannon, senior fellow at the Walton Family Foundation, where she supports a growing body of research on how teachers and students are using AI.
Measuring how teachers adopt AI is not straightforward. Joanna says the figures shift depending on what is asked and how AI is defined. The widely reported time savings of using AI come with significant caveats. And although AI can remove certain biases from grading, it introduces fresh risks of its own. When is reaching for AI appropriate, and when does it feel misplaced?
Bio
Joanna Cannon is a senior fellow with the Walton Family Foundation, where she works with the K-12 education team, including on a portfolio of research on how AI is being adopted by teachers and students. Joanna holds a PhD in developmental psychology from Columbia University and completed a postdoctoral fellowship at the University of Chicago. Previously, she was chief operating officer at the education non-profit PENCIL and chief strategic officer for the Division of Talent, Labor and Innovation at the New York City Department of Education.
Links
Walton Family Foundation, Education Program
Gallup and Walton Family Foundation research on teacher AI use and time savings

Join us on social media: 
BOLD (@BOLD_insights), Libby Hills (@Libbylhhills) and Owen Henkel (@owen_henkel)
Listen to all episodes of EdTechnical here: https://boldscience.org/podcasts/edtechnical/
Subscribe to BOLD’s newsletter: https://bold.expert/newsletter
Stay up to date with all the latest research on child development and learning: https://bold.expert
Credits: Sarah Myles for production support; Josie Hills for graphic design; Pia Schmoeckel for content production.

In this episode of EdTechnical, Owen reports back from the Festival of Learning in South Korea, the annual gathering that brings together academics in AI, education, learning sciences, and human-computer interaction. Owen takes the pulse of the field through four conversations at the Festival. He speaks with a Cornell professor, a funder at the Gates Foundation, an early-stage startup founder, and a researcher at Google DeepMind.
The community has stopped asking simply whether generative AI is good or bad for learning, and started asking where to use it, where not to, and how to build around it.

Guests
Rene Kizilcec is a Professor at the Department of Computing and Information Science at Cornell University and a general chair of the 2026 Festival of Learning. His work looks at how to use technology effectively and responsibly to support student learning.
Bryan Richardson works on the K12 education program at the Gates Foundation, where he focuses on helping AI models get better at education. He is part of a newly launched, pooled AI infrastructure fund backing open source tools and benchmarks for education.
Maya Bialik is founder of QuestionWell, an AI tool that helps teachers create and refine instructional materials. She is also a PhD student at Boston University, and author of Irreplaceable: How AI Changes Everything (and Nothing) in Teaching and Learning. 
Kevin McKee is a research scientist on the AI for Education team at Google DeepMind, where he leads assessments and evaluations of Gemini and other large language models in real world educational settings.

Links
The Festival of Learning 2026
Owen’s research on LLMs’ ability to interpret and grade handwritten student work 
Further reading from the conversations:
Catching The Correct Answer Trap by Moiz Imran and Sahan Bulathwela
From Untamed Black Box to Interpretable Pedagogical Orchestration by Nizam Kadir
Research by QuestionWell
Eedi and Google DeepMind math tutoring research in the UK
Gemini Guided Learning study with Fab AI in Sierra Leone

Join us on social media: 
BOLD (@BOLD_insights), Libby Hills (@Libbylhhills) and Owen Henkel (@owen_henkel)
Listen to all episodes of EdTechnical here: https://boldscience.org/podcasts/edtechnical/
Subscribe to BOLD’s newsletter: https://bold.expert/newsletter
Stay up to date with all the latest research on child development and learning: https://bold.expert
Credits: Sarah Myles for production support; Josie Hills for graphic design; Pia Schmoeckel for content production.

In this episode of EdTechnical, Libby and Owen speak with Rebecca Winthrop, senior fellow at the Brookings Institution, where she directs the Center for Universal Education. Rebecca and her team recently ran a ‘pre-mortem’ on generative AI in education: rather than waiting years for the evidence to settle, they conducted a yearlong study on the risks it poses to students, and what we can do now to prevent them while making the most of the benefits.
Drawing on focus groups, a thorough review of studies, and a forecasting panel, the report paints a striking picture: as things stand, the risks of generative AI for children's learning and development outweigh the benefits. Rebecca explains the risks are foundational, undermining the trust, independent thinking, and resilience that children need in order to benefit from AI. A key distinction runs through the conversation: risks tend to come from wide AI use, where children have open-ended, unscaffolded conversations with tools not designed for them, while benefits come from narrow, intentional use integrated into teaching with vetted content.
Bio
Rebecca Winthrop is a senior fellow at the Brookings Institution and director of its Center for Universal Education, where she leads the Brookings Global Task Force on AI in Education. She advises governments, policymakers and international institutions on education and previously led education work for displaced and conflict-affected communities. She is a co-author of the report A New Direction for Students in an AI World: Prosper, Prepare, Protect.
Links
Research Report by the Brookings Institution’s Center for Universal Education:
A New Direction for Students in an AI World: Prosper, Prepare, Protect
Brookings Institution Center for Universal Education: https://www.brookings.edu/topics/education-2/

Join us on social media: 
BOLD (@BOLD_insights), Libby Hills (@Libbylhhills) and Owen Henkel (@owen_henkel)
Listen to all episodes of EdTechnical here: https://boldscience.org/podcasts/edtechnical/
Subscribe to BOLD’s newsletter: https://bold.expert/newsletter
Stay up to date with all the latest research on child development and learning: https://bold.expert
Credits: Sarah Myles for production support; Josie Hills for graphic design; Pia Schmoeckel for content production.