Can AI really accelerate student learning by years?

Find out more

Highlights

- It's a fascinating look at where the rubber is really meeting the road.
- It’s an example of enhancement, not replacement, allowing students to access personalized support that might otherwise be unavailable.
- The authors then pivot to a study conducted in Northern Italy, which focused on the impact on educators.
- And critically, they didn't just sit back; they *directly reallocated* that time to 1:1 student mentorship, motivational support, and emotional support.
- In India, they're expanding the Google AI Educator Series, offering practical, mobile-first training customized to the unique needs of Indian educators.

Transcript

SPEAKER_00 0:00

If this episode makes you think, please let us know in the comments and support us by subscribing and leaving a review. Thank you. Today we are exploring a really interesting piece from the Google blog titled Measuring the Impact of AI on Teaching and Learning. It was co-authored by Chris Phillips, who's the general manager for Geo and Education at Google, and Laila Ibrahim, the Chief AI Readiness Officer at Google DeepMind. Essentially what they're doing in this piece is sharing the early results of two new impact studies that look at how their Gemini models are being used in real classrooms, specifically in Sierra Leone and Italy, and then they connect that to some new global initiatives for building AI literacy amongst educators. It's a fascinating look at where the rubber is really meeting the road. Now what really caught my eye first was the study they described from Sierra Leone. This wasn't just a casual observation, it was an eight-week pre-registered randomized controlled trial, an RCT, which is a pretty robust way to measure impact. They worked with Fab AI and local teachers, randomly assigning 48 math classrooms, nearly 1,800 junior secondary math students in grades 7 and 8, to either use a tool called Guided Learning, which is powered by Gemini, or just continue with their regular classwork. And here's where it gets really interesting. The students who used the AI-powered guided learning significantly improved their mastery of foundational topics like fractions, exponents, and prime numbers. We're talking about an increase in their scores on externally validated assessments by 0.26 standard deviations. Now for those of us who aren't steeped in educational research stats, they put it into really clear terms. This is equivalent to roughly 1.2 to 1.7 years of typical learning progress in low and middle income countries. That's a huge leap, isn't it? Think about what that means for a student who might be struggling with these core mathematical concepts. AI is stepping in, augmenting the teacher's efforts and really helping to bridge some significant learning gaps. It's an example of enhancement, not replacement, allowing students to access personalized support that might otherwise be unavailable. What else really struck me about the Sierra Leone study was the engagement. We often hear about how EdTech tools struggle with adoption, but in this trial, engagement was incredibly high. They found that 69% of the guided learning students reached an intended 12-hour usage threshold, with the average across schools being 15 hours of Gemini use. And the students who hit that recommended usage they saw even bigger gains, moving up to a 0.38 standard deviation increase. That effectively moves an average student from the middle of their class into the top third. When I hear that, I immediately think about the middle 80%. Those students who often don't get the specialized attention of the top achievers or the intensive intervention of those at the very bottom. This AI tool is clearly serving that often invisible majority, giving them tailored practice and pushing them forward in a really meaningful way. It shows that when the technology is purposeful and grounded in learning science, and when teachers are supported in its deployment, students embrace it. But it's not just about students, is it? The authors then pivot to a study conducted in northern Italy, which focused on the impact on educators. This one was across the Don Bosco School Network and involves surveys, focus groups, and over 560 detailed teaching activities, with 700 educators and 9,000 students, spanning everything from primary to upper secondary to vocational college training. Now this is the bit that really got me thinking about how we redefine the teacher's role. Educators were using Gemini for education to assist with content creation and scaffolding. The result? They successfully personalized their learning materials, and teachers reported that an impressive 80-99% of students in each class successfully mastered their planned lesson skills. That's remarkable from calculating the geometry of a parabola to writing Java code. It suggests that AI isn't just about speeding things up, it's about enabling a level of differentiation and individualized support that's incredibly difficult for one teacher to achieve alone, even with the best intentions. But here's the real kicker for me, and this is where the human in the loop and outsource the doing, not the thinking. Philosophies really come alive. Teachers reported a 70% reduction in time spent on administrative tasks. 70%. Can you even imagine what that means for a teacher's workload and well being? And critically, they didn't just sit back, they directly reallocated that time to 1.1 student mentorship, motivational support and emotional support. This is exactly what we're aiming for, isn't it? We talk about aiming high in AI adoption, not just for efficiency, but for deeper impact. AI is helping us hold the complexity of all those administrative tasks so we have capacity for creativity, and more importantly, for connection. This allows teachers to lean into those uniquely human domains that AI cannot touch care, judgment, relationship, imagination. Machines can compute, they cannot wonder, they cannot care, but when they free up human time, those human qualities can truly flourish. The piece then moves into a crucial area, AI literacy. The authors acknowledge that while teachers are eager to use these tools, it can be overwhelming to master new technology at the level required to guide their students. This is so true. We often label teachers as resistant to change, but more often they just need time and space and the right kind of support. Give them that, and they become the best drivers of innovation. Google is launching new initiatives to address this, focusing on India and the African Union Commission. In India they're expanding the Google AI Educator series, offering practical mobile first training customized to the unique needs of Indian educators. And this isn't a one size fits all approach. They're localizing the training into six languages in its first year. That's a powerful commitment to equity and accessibility, ensuring that AI literacy is not an afterthought, but a foundation for widespread adoption. It goes back to tailoring the solution to the institutional context, rather than just dropping a generic tool in and expecting magic. Similarly, their partnership with the African Union Commission aims to support AI literacy across all 55 member states, starting by bringing Gemini for Education and Notebook LM to university students and faculty. Again, it's about providing administrators and faculty with no-cost product onboarding and training playbooks, helping university faculty manage administrative tasks so they can focus on high-level instruction. It's that exact same principle from the Italy study, but scaled up to a continental level, free up the doing, so the thinking, the teaching, and the caring can take center stage. What this article really underscores for me is that the conversation about AI in education isn't just theoretical anymore. We're seeing rigorous studies demonstrating tangible benefits for student learning and teacher capacity. But crucially it's not just about the tools themselves, it's about the thoughtful, deliberate strategies for empowering educators through AI literacy. It's about customizing the approach, understanding existing friction points, and always starting with educational purpose over technological capability. It's an evolution, not a revolution, but one that is steadily and significantly reshaping what's possible in our classrooms. That's all for today. Thanks for listening.