
Sign up to save your podcasts
Or


In this short, Owen and Libby discuss a recent World Bank blog post about a study in Nigeria that evaluated the impact of Microsoft Copilot (powered by ChatGPT) on student learning outcomes. In a six-week after school programme, students were supported to use Copilot. The full study hasn’t been published yet but the blog post reports “overwhelmingly positive effects on learning outcomes”. It reports that the learning improvement over the six-week programme was equivalent to nearly two years of typical learning. Owen has a few questions about this…
Link
World Bank blog post ‘From chalkboards to chatbots: Transforming learning in Nigeria, one prompt at a time’
Join us on social media:
Credits: Sarah Myles for production support; Josie Hills for graphic design; Pia Schmoeckel for content production.
In this episode, Libby and Owen interview Mike Frank, Professor at Stanford University and leading expert in child development. This episode has a different angle to the others, as it is more about AI as a scientific instrument rather than as a tool for learning. Libby and Owen have a fascinating discussion with Mike about language acquisition and what we can learn about language learning from large language models. Mike explains some of the differences between how large language models develop an understanding of human language versus how babies do this.
There are some big questions touched on here, including how much of the full human experience it’s possible to capture in data. Libby and Owen also make excellent use of Mike’s valuable time by asking for his expert view on why infants find unboxing videos - videos of other children opening gifts - so addictive.
Links
Join us on social media:
Credits: Sarah Myles for production support; Josie Hills for graphic design; Pia Schmoeckel for content production.
In this short, Libby and Owen discuss a hot-off-the-press study that is one of the first to test how ChatGPT impacts the time science teachers spend on lesson preparation. The TLDR is that teachers who used ChatGPT, with a guide, spent 31% less time preparing lessons - that’s 25.3 minutes per week on average. This very promising result points to the potential for ChatGPT and similar generative AI tools to help teachers with their workload. However we encourage you to dig into the summary and report to go beyond the headline result (after listening to this episode) - this is a rich and rigorous study with lots of other interesting findings!
Links
Join us on social media:
Credits: Sarah Myles for production support; Josie Hills for graphic design; Pia Schmoeckel for content production.
This episode is the second in our three-part mini-series with Google, where we find out how one of the world’s largest tech companies developed a family of large language models specifically for education, called LearnLM. This instalment focuses on the technical and conceptual groundwork behind LearnLM. Libby and Owen speak to three expert guests from across Google, including DeepMind, who are heavily involved in developing LearnLM.
One of the problems with out-of-the-box large language models is that they’re designed to be helpful assistants, not teachers. Google was interested in developing a large language model better suited to educational tasks, that others might use as a starting point for education products. In this episode, members of the Google team talk about how they approached this, and why some of the subtleties of good teaching makes this an especially tricky undertaking!
They describe the under-the-hood processes that turn a generic large language model into something more attuned to educational needs. Libby and Owen explore how Google’s teams approached fine-tuning to equip LearnLM with pedagogical behaviours that can’t be achieved by prompt engineering alone. This episode offers a rare look at the rigorous, iterative, and multidisciplinary effort it takes to reshape a general-purpose AI into a tool that has the potential to support learning.
Stay tuned for our next episode in this mini-series, where Libby and Owen take a step back and look at how to define tutoring and assess the extent to which an AI tool is delivering.
Team biographies
Muktha Ananda is Engineering leader, Learning and Education @Google. Muktha has applied AI to a variety of domains such as gaming, search, social/professional networks and online advertisement and most recently education and learning. At Google Muktha’s team builds horizontal AI technologies for learning which can be used across surfaces like Search, Gemini, Classroom, and YouTube. Muktha also works on Gemini Learning.
Markus Kunesch is a Staff Research Engineer at Google DeepMind and tech lead of the AI for Education research programme. His work is focused on generative AI, AI for Education, and AI ethics, with a particular interest in translating social science research into new evaluations and modeling approaches. Before embarking on AI research, Markus completed a PhD in black hole physics.
Irina Jurenka is a Research Lead at Google DeepMind, where she works with a multidisciplinary team of research scientists and engineers to advance Generative AI capabilities towards the goal of making quality education more universally accessible. Before joining DeepMind, Irina was a British Psychological Society Undergraduate Award winner for her achievements as an Experimental Psychology student at Westminster University. This was followed by a DPhil at the Oxford Center for Computational Neuroscience and Artificial Intelligence.
Link
The LearnLM API
Join us on social media:
Credits: Sarah Myles for production support; Josie Hills for graphic design; Pia Schmoeckel for content production.
In this episode, Owen and Libby chat about AI tutoring with guests, Ben Kornell, Managing Partner at Common Sense Growth Fund, and Alex Sarlin, a veteran in the edtech industry. Both co-founded Edtech Insiders, a leading newsletter and podcast covering the growing Edtech industry.
Ben and Alex differentiate between AI-powered search and true AI tutoring, and discuss trends like AI-enhanced human tutors, hybrid models, and fully autonomous AI bots. The conversation highlights the need for AI to integrate- and learn from traditional education in developing key elements, such as targeting the right zone of proximal development. Human tutors have the ability to sense motivation and frustration, helping students through the more challenging parts of learning. Emerging technologies are now using facial and physical cues to gauge engagement, proving valuable as nudges for AI tutors or human instructors to boost motivation or adjust level of content.
They also address ethical and political risks, such as biased responses and dependency issues. With exciting developments on the horizon, the episode explores the at times seemingly sci-fi-like future of AI tutoring!
Guest bios:
Join us on social media:
Credits: Sarah Myles for production support; Josie Hills for graphic design; Pia Schmoeckel for content production.
This episode is the first of a three part mini-series with Google. There is a lot of interest in how big tech companies are engaging in AI and education and what their future plans are - in this mini-series, hear the latest directly from Google.
The genesis of this mini-series was a short Ed-Technical episode from earlier this year. Libby and Owen discussed a paper Google released about the work they had done to fine-tune a LLM called LearnLM to make it more useful for education. This work was motivated by a realisation that some of the core behaviours of LLMs (helpfulness, sycophancy) aren’t aligned with what’s valuable from a learning perspective, and prompting can only go so far.
This first episode focuses on how Google is integrating LearnLM’s capabilities into existing Google products like YouTube and new products like LearnAbout. The next episode in the mini-series will focus on LLMs and tutoring, and the final episode will be a more technical episode on the development of LearnLM.
We had a chance to talk to a number of folks across a range of teams, including LearnX, Google Research and DeepMind. There was too much great content to squeeze into three episodes but all full interviews will be up on our YouTube channel.
In this episode we include excerpts from interviews with four members of the team.
Rob Wong is the Product Lead for LearnX, a team within Google that builds learning features on Search, YouTube, and Gemini chat, and also works on LearnLM in partnership with Google Research and Google DeepMind.
Julia Wilkowski leads a pedagogy team at Google. Her team collaborates with Google product teams to apply learning science principles and teaching best practices.
Markus Kunesch is a Staff Research Engineer at Google DeepMind and tech lead of the AI for Education research programme. His work is focused on generative AI, AI for education, and AI ethics, with a particular interest in translating social science research into new evaluations and modeling approaches.
Angie Mac McAllister, PhD is a Group Product Manager at Google with a vision: to make a personal AI tutor available to everyone. Focused on developing learning features for Gemini, Mac combines 35 years of experience in education with cutting-edge AI to help students become better learners.
Links:
Join us on social media:
Credits: Sarah Myles for production support; Josie Hills for graphic design; Pia Schmoeckel for content production.
This episode, Owen and Libby speak to Ryan Baker, a leading expert in using big data to study learners and learning interactions with educational software. Ryan is a Professor in the Graduate School of Education at the University of Pennsylvania, and is Director of the Penn Center for Learning Analytics.
Ryan provides an overview of educational data mining (otherwise known as EDM) and explains how insights from EDM can help improve learner engagement and outcomes. Libby and Owen also explore the technical aspects of algorithmic bias with Ryan, discussing why it matters, how it is defined, and how it can be addressed from a technical perspective.
Links:
Join us on social media:
Credits: Sarah Myles for production support; Josie Hills for graphic design; Pia Schmoeckel for content production.
In this short episode, Libby and Owen discuss OpenAI’s new model for advanced reasoning, o1. They talk about its new capabilities and strengths, and what they think about its significance for education after an initial play around. They talk through the benefits of ‘think aloud’ versus ‘think before you speak’ approaches in education, and how this relates to o1.
Links:
Join us on social media:
Credits: Sarah Myles for production support; Josie Hills for graphic design; Pia Schmoeckel for content production.
In this episode, Libby and Owen are joined by Craig Barton, Head of Education at Eedi and host of the Mr Barton Maths and Tips for Teachers podcasts, along with Simon Woodhead, Director of Research at Eedi. Together, they explore the world of educational misconceptions—what they are, why they matter and how AI and data science can help tackle them.
Links:
Join us on social media:
Credits: Sarah Myles for production support; Josie Hills for graphic design; Pia Schmoeckel for content production.
In this short, Libby and Owen discuss recent research from Anthropic looking at sycophancy – the tendency to agree with users – in large language models (LLMs), and key research from educational psychology about how important feedback is for learning. Libby and Owen connect the two papers and explore why sycophancy is especially a problem when it comes to using LLMs for educational purposes.
Links:
Join us on social media:
Credits: Sarah Myles for production support; Josie Hills for graphic design; Pia Schmoeckel for content production.
From the publisher's feed
Hosted by EdTechnical co-founders Libby Hills (CEO) and Owen Henkel (Research Director), the EdTechnical podcast explores AI in education through a research-grounded…
Each episode, Libby and Owen ask experts to help educators sift the useful insights from the AI hype. They ask questions like: how does this actually help students and teachers? What do we actually know about this technology, and what is just speculation? And (importantly!) when we say AI, what are we actually talking about?
Beyond the podcast, EdTechnical also invests in promising AI edtech companies and conducts applied research to inform real-world product and investment decisions.