At Google I/O, VP Chris Phillips discussed new AI research and education announcements. A randomized control trial in Sierra Leone demonstrated that Gemini tutoring accelerated English learning by 1.2-1.7 years, marking a pivotal evidence base for AI in education. Teachers using AI tools reported up to 70% time savings on administrative tasks, which they reinvested in student interaction. Google launched an AI Education Series with short training modules for teachers, rolling out in the US, India (in six languages), and across 55 African Union states. Phillips highlighted that AI should be teacher-led to enhance, not replace, educators, and to foster critical thinking and collaboration. Key announcements for schools include Gemini-powered study guides and adaptive learning tools. Phillips noted that device strategy matters less than effective teacher guidance—AI can succeed in various classroom settings. To mitigate risks of student dependency, Google embeds learning science into its models to promote active learning and manage cognitive load. The long-term vision is a "symbiotic" relationship where AI helps every student achieve their potential, enabling teachers to build stronger relationships and students to master subjects with deeper engagement.
I am at the Shoreline Ampitheater in Mountain View Silicon Valley at Google I/O. This is Google's flagship annual event. A festival of everything Google and as you can imagine AI was a huge part of the event. After the two-hour keynote from the CEO Sundar Pichai, I got to interview Chris Phillips. Chris is the VP and General Manager of GU in Education at Google. That means he oversees Google Maps, Ways, Earth, Street View and Google for Education. We chatted about some new AI and education research that Google just released. And what the major announcements from the keynote might mean for education. I hope you enjoy it. Let's jump in. Chris, you've published the first major randomised control trial of Gemini and Astridance today. 1.2 to 1.7 years of learning process and Sierra Leone. Critics have argued for two years at the evidence base for AI and education in STEM. Is this the moment that the argument changes and what's the obligation on smooths now? Do you think it's an opportunity? Yeah, it's a great question. We think this is just the beginning. This study we're really proud of, done very diligently, is showing evidence that these tools with the right application and a teacher-led environment can really create great outcomes for students. And the guided learning product and experience is super-engaging for the students. So you see the students leaning in to learn more about the topic. And the acceleration in their learning is something we're really excited about. I think what this is though, when I say it just the beginning, is that we feel the responsibility to continue to partner, work with educators around the world to continue to show evidence that when applied correctly and responsibly, this can deliver great outcomes for both teachers and students in environments all around the world. Mason, I know that Don was supposed to be an 8-inch-inch, about 70% reduction in argument times, 70% reallocation to want to want to show. What did teachers actually have to do differently to get that result? Because my audience is teachers, so that figure of 70% is going to stick with them. So what did they have to do to get that result? And what's the risk that that school's here? AI saves time and never reinvest those hours where you did? So we think that in this moment right now, getting teachers in the lead, getting educators trained on how to use AI in practical ways, there is such a huge opportunity. And it can deliver real time savings. We also had a study we did in Northern Ireland that showed real time savings and then their recent study in Italy as well. So we're excited to continue to have teachers share with other teachers, the success they're having. Even launched a Google Education Series, which is like bite size, small training modules, that take all kinds of different tasks and activities that teachers do day to day, and showing how they can use AI to help them. And there's so much administrative time spent by teachers, so many nights and weekends spent. And this is a great opportunity. We think to give time back to educators so they can spend more that time with the students. You know, we don't believe AI is going to replace teachers. We believe this is an opportunity to enhance, help them so they can have that human connection with students and spend more of the one-on-one time and create interesting and engaging lesson plans and use different formats. That otherwise would have been like really time-consuming and hard to do. The AI makes those things easier and we see the students responding really well to it too. Mason, we've just watched the keynote by Sundar and all your guys from the stage of everything that was announced today. Which single announcement do you think a principal, let's say K12 principal, should be paying attention to the most? The presentation that Josh did around how to use Gemini for study guides and what we're going to do in a test prep and even the guided learning, those are really great applications of AI that every principal, every teacher and we believe every student will love using. Think about the intersection where the use of AI truly helps every student and can adapt to their learning style and their pace but is still led by the teacher. So whatever that lesson is or whatever you're studying that test you want to do well on, based on the curriculum, based on the content that the teacher or the administration wants to use, these tools can help students catch up, get ahead, advance forward and we really think that's a win-win for educators and students. You mentioned the AI education series and you're localising that into six Indian languages and partnering with the African Union across 55 member states. Is the centre of gravity for AI in education shifted away from the US and Europe and what is happening for schools in the US and the UK where I'm from that might think they're out in front of us? We have a worldwide focus so we believe what we want to build for the education series ultimately rolls out around the world. We announced launching that in the US, that series is kicking off for educators which really cool about it is it's by size training, we're partnering with ISD but it's also teachers training teachers which we think is really important and we want to help provide helpful technology but educators in the lead is really important. And so by doing that whether it's in Africa or India and the US we have ambitions to do that worldwide and we'll localise it in the right languages with the right content, with the right partners because as we go around the world it's about how we partner with the communities and the educators in those markets. We're not going to go out of the loan. We really think it will land much better when it's teacher led with the local experts. I think this is a lot of concern from a lot of educators around the potential of AI to make students dependents on the AI tools and not do their own thinking. I suppose as one of the many people responsible for putting a powerful AI in front of students who are well. What are you guys doing to kind of mitigate against those potential risks? Yeah one big thing we're doing to mitigate the risks is to get teachers and educators comfortable with the UCDI. One thing we've seen over and over is when the teacher leads with here's the appropriate UCDI for their students. Here's how we're going to change how we're going to assess what you learn. Here's how we want you to explain what you understood and even drive a deeper learning of a topic. That's when we see really great things happening. We really think the key here is to get educators whether it's faculty members in higher ed or teachers in K-12 comfortable with the use of the technology to teach the lessons that they're expecting. We see students loving the fact that it can actually adapt to their learning style, whether it's an audio format or a video based on what the teacher is instructing. We see a lot of students getting even more active in their learning. They're actually engaging more deeply. The study is we're showing that they're not just seeking the quick answer. They're actually wanting to dig in and master a topic. What's really important this moment is to help the teachers and the educators because this is here to help them not replace. We still believe strongly that these tools can help the educators and that's really important to avoid those risks that we just talked about. Yeah, it's interesting because the people I own, Susan, sorry, we're through I use an AI well. They're actually going on cognitive overloads. They're not giving up cognitive, but they like to say they're learning a new way. It's exciting. The biggest gains in Sierra, they came from students who had 15 hours of gym at English. Should students be thinking about AI, the way they think about reading minutes, so like a minute, you create exposure, are you comfortable with the ability to effectively set and establish that? Just trying to take the researcher and turn it into something meaningful for educators around the world. Yeah, we don't intend to set any standard of how much use. I think part of setting a measuring that usage and the dosage, so to speak, was to understand at what point do you see certain measured results? With that said, we think it's really going to be based on the teacher, the students, the environment, the topic, the subject you're teaching, it's really going to be different. And frankly, it doesn't need to be huge amounts of usage to deliver big results. You can take a complex topic and for a student to have to break it into smaller pieces, or for a teacher to be able to build a lesson plan that's got three versions for the different kinds of learning happening in their classroom, those are things that can unlock big benefits that don't equal lots of time spent per se. So it's really about, that's why that training on different ways they can use the tools. And over time, the teachers will see how to best apply it to continue to have that human connection. And that's really important. This is not AI taking up the space in the classroom. This is back to the teacher and the student relationship, and getting excited about the topic you're learning and using engaging formats. And so we're not intending to set any standard on that. We just want to help educators feel comfortable with the tools and trust and build trust so that they can then guide how it's used. The best decision makers on how much time should be spent, should be driven by the administrators, the educators themselves. They see a little try work in classrooms where
I suppose it's probably a very different to what we see in the US, UK classroom with shared devices, partnership connectivity. I suppose it's also to take a lot of more in classrooms that might have want to go on Chromebooks and so on. Does this research tell us that the device strategy of the last decade matters less than people thought? What should that change about how soon to spend the next infrastructure? Well, technology by itself is not going to deliver the learning outcomes. It's a tool. It's there in support. We think that multiple environments can have positive outcomes if they're managed effectively. If you're in an environment where there's a one-to-one device, we still are building tools where the teacher controls is the device on off, how it's used, the portion of the class words. We think there's a lot of controls that we can be built into that. We built features to do that. For you're in a classroom where there's one device, and it's a shared device, and that device is being used. I was in a classroom where they used notebook alarm to take a storybook, translate it, build a picture view of it, and we're teaching it in multiple languages. All the students were reading it at the same time, and the teacher was able to look at all the students directly. It was amazing. I think the answer is going to be different in different places around the world, where the access to the technology is different. But in every case, it's got to be the teachers and the educators leading how it's used, how much time on what curriculum. We're here to support that. In those situations, I think there'll be positive outcomes when it's led that way. That's why it's so important as we roll this out around the world. We need to be partnering closely with all the local entities because they will understand the right environment. In some places in the world, it's going to be a mobile device or a smart word or software. All of that still is assisting the teacher in that moment. That's really, really important. Sometimes, as somebody who tries to translate what's going on, it's going to invite for teachers and classrooms. It feels like it's at multiple stages out of the air. There's a constant waste between Google, Google, Google, and all of the AI and fraud that you might for some of. Do you, I suppose, as somebody who works from the inside of Google within the education space, does it feel like from the inside? Does it feel like you guys are in a race to get your suite of tools in front of students or GC down a different age? Well, Google has always been focused on making information, university accessible and useful. Education is such a core part of that. We've had educational cards for 20 years. We're not new to having our products progressively get better and being in the classroom in support of the teachers and learning. In this moment where this introduction of gender to the AI and how large language models are used and the progression, it's super exciting moment because it's unlocking all kinds of new use cases. We see it as our responsibility, though, to be really thoughtful and careful to focus on genuinely driving great learning outcomes. That's why we've taught our model how to embed learning science into the actual way that the model behaves in an educational academic use case. How it can drive active learning, how it can adapt to the learner themselves, how it can manage the cognitive overload that could happen so that we can break topics down and help people really digest. Our guiding principle here is to deliver really great learning outcomes and be trusted by educators and students in this moment. It's an exciting time where there's a lot of investment coming into this space. We feel confident that if we do the right thing and we build trust and deliver real learning outcomes, that good things will happen. I suppose let's try and do what we probably shouldn't in that is get the crystal ball out. I only say this because I was really struck by the dentist at this morning about how he thinks we will look back and recognize that we were in the footfields of the symbiarit. That chord really struck me. If we're in the foothands of the symbiarity, first of all, for you, because I suppose there's so many different definitions of the symbiarity, but I suppose for you in Google what is that symbiarity? Secondly, what does education, I'll be education system, traditional student, that's right, when we have a convergence of AI, chalo under the term, listening to your lousy. Education is so critical to helping people achieve their full potential. And with AI, I'm really hoping that we're at the foothands, that will advance upward into teachers and educators having that confidence on how to use it. That means transforming and building the content they use to teach. That means how the students will actually be building critical thinking skills and communication and collaboration and resilience while they're learning subjects. Not as a separate thing to learn, but integrated in. So I'm hoping that as we continue to stay connected on this path, that we're seeing more and more educators telling stories of student success. Students that might have been falling behind, now can catch up and get master a topic that otherwise was difficult for them. Teachers that have large classrooms and have tons of administrative overhead are now having that one-on-one time with their students. And back to the relationship building that we see students coming out, not just learning about history and science and reading. All of these things are critical, but they're doing that with new skills applied today. So they can take a topic they learned about and debate it and argue both sides of it. And that skill match with learning the topics is such an exciting thing that these tools can help unlock. So that singularity is, I think, really helping advance what effectively the hope that all teachers can have that relationship with each student. Each student can find the things that they love and want to master and catch up on the things that they're struggling with. And some of the studies are showing kids that maybe weren't as interested in math, now love math, or maybe not as interested in that history lesson. And now we're able to make it their own and and and be able to talk about both sides of that in a deeper way than maybe we traditionally taught. And that is an exciting, exciting time for what I think we're just at the beginning phases of. My full article about the announcements at Google I/O and what they mean for education can be found in tomorrow's edition of the newsletter. Sign up at the a ieducator.io to get it in your mailbox tomorrow.
Podcast Summary
Key Points:
Google released a randomized control trial in Sierra Leone showing AI tutoring (Gemini) led to 1.2-1.7 years of learning gains in English.
The study emphasizes teacher-led AI use; teachers saw up to 70% reduction in administrative tasks, allowing more one-on-one student time.
Google launched an AI Education Series with bite-sized training modules, localized into six Indian languages and partnering with the African Union.
Key announcements for K-12 include Gemini for study guides, test prep, and guided learning that adapts to student pace and style.
Google stresses AI as a tool to enhance, not replace, teachers, with focus on human connection and critical thinking.
Device strategy matters less than teacher guidance; AI can work in one-to-one or shared-device classrooms.
Google embeds learning science into AI to manage cognitive load, promote active learning, and avoid student dependency.
The long-term vision is a "symbiotic" future where AI helps every student achieve potential and teachers build deeper relationships.
Summary:
At Google I/O, VP Chris Phillips discussed new AI research and education announcements. 7 years, marking a pivotal evidence base for AI in education. Teachers using AI tools reported up to 70% time savings on administrative tasks, which they reinvested in student interaction.
Google launched an AI Education Series with short training modules for teachers, rolling out in the US, India (in six languages), and across 55 African Union states. Phillips highlighted that AI should be teacher-led to enhance, not replace, educators, and to foster critical thinking and collaboration. Key announcements for schools include Gemini-powered study guides and adaptive learning tools.
Phillips noted that device strategy matters less than effective teacher guidance—AI can succeed in various classroom settings. To mitigate risks of student dependency, Google embeds learning science into its models to promote active learning and manage cognitive load. The long-term vision is a "symbiotic" relationship where AI helps every student achieve their potential, enabling teachers to build stronger relationships and students to master subjects with deeper engagement.
FAQs
The trial showed that using Gemini and guided learning with a teacher-led environment resulted in 1.2 to 1.7 years of learning progress in English, demonstrating significant acceleration in student outcomes.
Teachers can save time by using AI for administrative tasks, with studies showing up to 70% reduction in time spent on such tasks, allowing them to reinvest that time in one-on-one student interaction.
The Google Education Series is a set of bite-sized training modules that show teachers how to use AI for daily tasks, helping them save time and integrate AI effectively in the classroom.
The use of Gemini for study guides, test prep, and guided learning is key, as it adapts to each student's learning style while keeping the teacher in the lead, helping students catch up and advance.
Google focuses on training teachers to lead AI use, emphasizing assessment changes and deeper learning, so students engage more actively and seek to master topics rather than just get quick answers.
No, Google does not set a standard usage amount; the dosage depends on the teacher, students, and subject, and even small amounts of AI use can unlock significant learning gains.
Chat with AI
Loading...
Pro features
Go deeper with this episode
Unlock creator-grade tools that turn any transcript into show notes and subtitle files.