New Year, New Ideas with Google Part 2: Benedict Gomes and Maureen Heymans on AI, Learning, and Sustainability

Benedict Gomes, SVP of Learning & Sustainability at Google, leads efforts to transform education and tackle global climate challenges. His organization develops tools like Google Classroom, ReadAlong, and YouTube-based learning platforms, empowering educators and learners worldwide. Ben also drives sustainability initiatives, including decarbonization and creating global environmental data solutions.

Maureen Heymans, VP of Engineering, Learning & Sustainability at Google, oversees AI integration into tools like Search, YouTube, and Classroom to democratize access to quality education. She collaborates across teams to pioneer personalized learning technologies while leading sustainability initiatives that harness AI to address climate challenges and drive impactful global solutions.

💡 5 Things You’ll Learn in This Episode:

1. How Google is revolutionizing learning with products like Google Classroom, ReadAlong, and Learn About.
2. The role of AI in creating personalized tutors for students and teaching assistants for educators.
3. Insights into Google’s efforts to make sustainability a core part of its mission through AI and partnerships.
4. How Google collaborates across teams to integrate pedagogy into AI tools like LearnLM.
5. The vision for equitable access to personalized education worldwide, powered by AI.

✨ Episode Highlights:

[00:02:18] Ben shares how AI is transforming education and Google’s vision for supporting learning globally.
[00:03:48] Discussing the importance of multimodal learning tools and products like NotebookLM and YouTube.
[00:19:24] Maureen reflects on how project-based learning and the pandemic inspired her focus on education.
[00:37:16] Explaining LearnX’s collaborative approach to driving learning innovations across Google’s ecosystem.
[00:46:16] Introducing Learn About, a new AI-driven conversational tool combining pedagogy with the power of the web.
[00:53:30] Envisioning a future where AI makes high-quality, personalized education accessible to billions globally.

😎 Stay updated with Edtech Insiders! 

• Follow our Podcast on:
  • Apple Podcasts
  • Spotify
  • Amazon Music
• Sign up for the Edtech Insiders newsletter.
• Follow Edtech Insiders on LinkedIn!

🎉 Presenting Sponsor:

This season of Edtech Insiders is once again brought to you by Tuck Advisors, the M&A firm for EdTech companies. Run by serial entrepreneurs with over 25 years of experience founding, investing in, and selling companies, Tuck believes you deserve M&A advisors who work as hard as you do.

Chapters

• 2:36 Ben shares how AI is transforming education and Google’s vision for supporting learning globally.
• 4:06 Discussing the importance of multimodal learning tools and products like NotebookLM and YouTube.
• 20:05 Maureen reflects on how project-based learning and the pandemic inspired her focus on education.
• 37:57 Explaining LearnX’s collaborative approach to driving learning innovations across Google’s ecosystem.
• 46:57 Introducing Learn About, a new AI-driven conversational tool combining pedagogy with the power of the web.
• 54:11 Envisioning a future where AI makes high-quality, personalized education accessible to billions globally.

Transcript

[00:00:00] Ben Gomes: if you look at the structure of the brain, if you think about human evolution, people have been reading in large numbers for a relatively short period of time. 

We've been listening to each other for a very long time, and we've been, had vision for an even longer period of time. So I think those are natural things. When you look at the structure of the, how the human mind is constructed, a huge amount of computation is given over. to those modalities. And so I think it's really interesting that for a while the only method by which people could absorb new information was through reading. 

Yeah, there was this phrase, you know, you have to learn to read in order to read to learn. And you know, a lot of students actually struggle with that and they have difficulty with reading. So I think If we can unlock new modalities by which people can learn, I think that would really enable a lot of learning in people who otherwise would not necessarily succeed at learning in today's world. 

[00:00:55] Maureen Heymans: So, yeah, I mean, first, if you think about learning, it's, it's really called to Google mission of organizing the world's information and make it university accessible, right? So no surprise that many teams across Google work on learning. It goes from learning for school, like classroom, but also many students coming to a product To learning for work with like a project like grow with Google, we provide credentials, but also cloud learning where you can acquire a new skill set for your job, but also learning for life, right? 

So many people come because they want to learn a new topic. They want to learn a new hobby. And of course, YouTube is an amazing platform. For this, 

[00:01:39] Alex Sarlin: welcome to EdTech Insiders, the top podcast covering the education technology industry, from funding rounds to impact to AI developments across early childhood, K 12, higher ed, and 

[00:01:51] Ben Kornell: work. You'll find it all here at EdTech Insiders. Remember to subscribe to the pod, check out our newsletter and also our event calendar. 

And to go deeper, check out EdTech Insiders Plus, where you can get premium content, access to our WhatsApp channel, Early access to events and back channel insights from Alex and Ben. Hope you enjoyed today's pod. 

[00:02:18] Alex Sarlin: Ben Gomes leads the learning and sustainability organization at Google and is an advisor to the Google search organization. He oversees the teams focused on harnessing technology to scale education around the world. The learning team supports educators and learners around the world with products including Google Classroom, Google Scholar, Google Arts and Culture, and an AI based tutoring app, Readalong. 

They are also creating academic and learning experiences on Google Search and YouTube. As one of Google's first principal engineers and an early Google Fellow, Ben also led the Google Search organization, which 

included Google Search, Google Assistant, and Google News for more than 20 years. Prior to joining Google, Ben earned his BS from Case Western University in Ohio and his PhD in computer science from UC Berkeley. 

He was born in Dar es Salaam, Tanzania, and raised in Bangalore, India. We're here with Ben Gomes, the SVP, Senior Vice President of Learning and Sustainability at Google, just coming off an incredible panel with all the different Google leads. Welcome to the podcast. Welcome to EdTech Insiders. 

[00:03:26] Ben Gomes: Great to be here. 

[00:03:27] Alex Sarlin: Yeah. So first off, you've made a point a couple of times recently that it really sticks with me, which is that reading is not the most natural modality for people to learn. In fact, it's sort of unnatural and that some of the different modalities that are coming out in tools like Notebook LM and, and YouTube may be unlocking some interesting things for AI. 

I'd love to hear you talk about that. 

[00:03:48] Ben Gomes: Yeah. You know, when you think about reading for many of us, it feels very natural because we learn to do it at a young age. But if you look at the structure of the brain, if you think about human evolution, people have been reading in large numbers for a relatively short period of time. 

We've been listening to each other for a very long time, and we've been, had vision for an even longer period of time. So I think those are natural things. When you look at the structure of the, how the human mind is constructed, a huge amount of computation is given over. to those modalities. And so I think it's really interesting that for a while the only method by which people could absorb new information was through reading. 

Yeah, there was this phrase, you know, you have to learn to read in order to read to learn. And you know, a lot of students actually struggle with that and they have difficulty with reading. So I think If we can unlock new modalities by which people can learn, I think that would really enable a lot of learning in people who otherwise would not necessarily succeed at learning in today's world. 

[00:04:50] Alex Sarlin: Yeah, and this is for adults and children alike, and people have been voting with their feet for a long time on this. The amount of reading has gone down and the amount of YouTube watching, the amount of interactive discussions has gone up, the amount of use of podcasts and audio has gone up. So. You're already seeing that signal in the market. 

So tell us about some of the Google products and how they are sort of addressing this modality shift. 

[00:05:13] Ben Gomes: Yeah, let me talk to you about like a couple of products. So there's basic technology that we built that handles voices and all really well. But one of the key products that where this started was a product called Illuminate, where Krishnabarath was actually the person who introduced me to Google, was really interested in consuming technical papers. 

And was, wanted to consume them on the fly in an audio way. And, you know, so he worked with researchers in DeepMind to actually figure out how to build this kind of a podcast like experience. And the same thing 

happened with NotebookLM, where writers were trying to figure out, like, how do I create a surface where I bring together all my materials? 

And then how do I absorb what's interesting in those materials? And so the same technology has manifested itself in more than one place and for multiple uses. And I can see this really being used in many, many different ways because it unlocks fundamentally this kind of modality of learning through this audio modality, which is just, I think, more natural for some people. 

[00:06:17] Alex Sarlin: Yeah. And more mobile in many ways for people that can do it while doing other things. 

[00:06:21] Ben Gomes: I mean, I listen to audiobooks when I'm in the car all the time, right? And it's very good. I think one thing we have to pay attention to in all of these things is how do we make sure that people retain and learn through audiobooks just as we would with textbooks. 

It's not enough in a book to just present the material. You actually have to provide ways and hooks for people to actually remember, retain, understand the material. And the same thing is going to be true of these audio modalities. And so we need to pay attention to all of those approaches in these modalities. 

But I don't think we should be prejudiced towards, you know, reading versus audio one way or the other. 

[00:06:54] Alex Sarlin: Yeah. And, you know, you're mentioning the assessment piece of the puzzle. It brings up a couple of things within the Google ecosystem. One is YouTube can now put sort of in video assessments that are created by AI. 

And the other is LearnLM, this very powerful, you know, model trained on pedagogical research, much of which is about assessment and how to use formative assessment for learning. I'm curious, you know, how you see Google's role in helping that assessment actually happen. So it's not just consumption of material, but actually assessing and learning and retaining. 

[00:07:24] Ben Gomes: Yeah, I think this is really interesting. When I think about YouTube, it's an amazing collection of content and these creators are fantastic. They really make many concepts more imputed. But you still need that extra step of making your process of interacting with the content more active. You still need to make sure that your brain actually engages and doesn't just passively absorb the content, but actually like understand it. 

So I think in that context, I think the solution will be working with creators to figure out what are the tools that we can provide that creators can use that engages their audiences so that, you know, we maintain an engaging experience. While still also making sure that people are learning more, that they're absorbing more of the material. 

And I think there's a tremendous potential there. There's an enormous library of great content. And now we have the potential to turn that into more active learning. Yes. As well as create more active content in the future. So I'm really excited about that. There are features being built in YouTube to actually allow you to ask a question about a video. 

Right? And that will allow you to begin down this journey of like making your interaction with the video much more active than it has been. There's a project that was called ShiftBot, which was a creator called Adam Schiff, who then can be present with you while you're learning to code. So it's again trying to make something that would have otherwise been just watching a video something part of an active experience. 

So I think there's a tremendous amount of potential there. There's another thing you're asking about that. Oh, uh, learn LM, how to learn LM. Yeah, right. And so in order for all of these things to work well, I think the fundamental technology has to be built solidly. And we need to keep making that fundamental technology better for these use cases. 

So with LearnLM, we are pushing the boundaries of what you can do with pedagogy in these models, right? In order for pedagogy to work well, the instruction following has to be accurate for that kind of pedagogy, where it doesn't give you an answer, where it coaches you in the right way, where it encourages you properly, but not excessively. 

That kind of getting the right tone for learning is really important for these tools are really to be useful to help you learn. So we're trying to build that into these fine tuned models, which we call LearnLM. But then as we begin to understand how to do that, we want to upstream that into Gemini at large as much as possible so that all our models really benefit from that usage. 

[00:09:43] Alex Sarlin: Yeah, that's a very exciting vision. And we've sort of talked on the podcast about how the learning layer is a little bit of a missing piece in the LLM puzzle, because learning models, the foundational models are really meant to Serve you right to give you answers to sort of solve your problems for you. 

But as we know, learning is not about somebody else solving your problems for you. So that that learning layer and what learn all I'm doing is incredibly exciting. I want to ask you about your history because you have a long tenure at Google. You started in something very different than education. Tell us about your Google history and what brought you to this sort of education specific type of role. 

[00:10:16] Ben Gomes: So I've been at Google a long time and I was in search for about 20 years and, you know, from the early days of search and, you know, search is a fascinating technical problem and focuses on language. But, you know, for me, I've always been interested in education and learning just because, you know, There's no way I would be where I am today if I'd not had gone to good school. 

If my mother was not a school teacher, all of those things were formative for me. And I kept going back to India to visit schools to try and see like, what kind of a role can technology play in schools? And so that is something that was an ongoing interest for me, even while I was working in search during vacations and other times. 

So after 20 years, I wanted to work on learning or education. And I thought about. Various ways to do that, and then I realized after conversations with many people that Google is a place where many of the products are really learning oriented, you know, our mission is to organize the world's information, making university accessible and useful. 

Well, one of the most useful uses of information is to transmit learning, right? Yes. And people are. Unsurprisingly, using our products to learn, whether it's search, whether it's YouTube, Classroom, Google Scholar, Google Arts and Culture, many of these things are used primarily or importantly for learning. 

So I was like, well, if I, I already have a background, what Google can do. Can I make use of my background here as well as learn more about learning, work with teachers and pedagogical experts to bring more of that learning into our product. And then when the tsunami of AI hit, you know, it became important to also think about, well, what new things can we build? 

Yes. And so we're thinking about new ways to explore the space. Many of you have seen Notebook Alan, which is an exploration of the space of like clearly has applications for learning. We're working on this lab experiment called LearnAbout, which again is trying to break up a complex topic and give you scaffolding during learning and providing these pedagogical bits in the context of learning to make you stop and think, to get you to be more actively engaged. 

And so that's just the beginning, right? So there's so much more that's possible there. 

[00:12:16] Alex Sarlin: Yes. You mentioned Google Arts and Culture and there's also Google Books. I mean, you know, Google's been trying to sort of chronicle the world's information for decades now. Yeah. But a lot of it. As you sort of mentioned, it's amazing, but it's sort of out there to be found, right? 

Somebody can go look inside a museum in the Prado if they are so inclined, or if they have a teacher sending them on a field trip, but it doesn't sort of come to them. And it feels like that's an incredible opportunity here. If somebody is searching for something or in their email, there's so many. entry points where people might find it interesting to access some of that information. 

I'm curious how you see that ecosystem playing sort of proactively pushing learning to people. 

[00:12:52] Ben Gomes: I think that's a really interesting question because I think, you know, it is true that many of our products, people come to us with a particular need. But I think if we can stimulate their curiosity in that moment, we can take them to deeper experiences where they learn more. 

I just saw, you mentioned Google Arts and Culture. I just saw this wonderful experiment that the team has done. which is giving you an audio guide when you're in a particular location virtually. Now, many of us are not going to get to visit many of these amazing locations. But to actually be there with something that's a guide based on exactly what you're looking at is really quite wonderful. 

It's like, this is really kind of incredible as an experience, right? So to get to experience things that you otherwise couldn't. So I think there is a challenge because so many new things have been created. How do we actually bring them to users at the right moment and in the right way? But you know, it's a challenge that we've been working on for a long while. 

And so I think I have confidence that we'll get there. It'll take us a little bit of time to figure it out. Because we want to make sure we make keeping the user experience really good. We're giving them what they actually want and not just what we think they want. So we have to work through this carefully, but I think there's a lot of potential to bringing many of these experiences to our products over time. 

[00:14:01] Alex Sarlin: Absolutely. I had a personal experience like that. I was in the Metropolitan Museum of Art in New York, standing at the Temple of Dender, and I suddenly said, I wonder what would happen if I asked AI, To explain this to me in real time. And it was like, not only could I explain it, it could tailor it to what I really wanted to know. 

And I realized, Oh, every piece of art in this whole museum, I could say, I'm looking at this piece and it'll tell me the history of it. I'll tell me about the artist. And it was like this little personal epiphany, but. I mean, everybody should have that ability. It's incredible. It's in the phone in your pocket, can do this for you already. 

[00:14:32] Ben Gomes: It is absolutely incredible. And especially when you think about the context of art, people from very different backgrounds that are people who have like studied art history in college. And there are people who have never been exposed to a certain kind of art and much later in life. And yet art can unlock wonderful things in our minds if approached the right way. 

So it's almost a perfect example of like, Art, bringing the thing that you want to learn in a personalized way to each individual based on what your interests are, what is it in your mind that could relate to this work of art, right? Because for many people, I think, see art as maybe an elite pursuit, but it shouldn't be. 

It really should be accessible to everybody. And it should be accessible in a way that helps them change their lives, because it can bring joy, it can bring insight, it can bring a whole host of things that you otherwise might not have access to. But that's where I think AI can play this role of bridging that gap. 

And I'm beginning to see the beginnings of that. 

[00:15:28] Alex Sarlin: Yeah, and AI is a sort of way to bridge the gap between an individual. And the content, you know, quote unquote, content out in the world, whether that's art or search results or YouTube videos, it's incredibly powerful because exactly as you say, it can personalize, it can sort of encourage transfer. 

You can say, I'm looking at this site because I want a new job. I'm looking at this site because I'm curious, looking at this site because I want to teach it to somebody else. You can bring your purpose, which is something that, you know, traditional. even search engines, as sophisticated as they were, could never assess why you were searching something. 

Now it can. 

[00:16:00] Ben Gomes: Right. I think there's this combination between your purpose and also serendipity. I think both things matter. 

[00:16:05] Maureen Heymans: Yeah. 

[00:16:06] Ben Gomes: When I think about a museum, you know, there's this amazing, the thing, the piece of work you wanted to go and see, And then all the things you happen to see around it that might spark your curiosity. 

And that's happened to me many times. I mean, I went to the Natural History Museum and I saw this large display of the Grand Canyon. I'd gone to see, I think, something about astronomy. I saw this display of the 

Grand Canyon and it figured what has been like a 10 year interest for me in geology. And I stayed with that one exhibit for like an hour. 

So it was just serendipity. But I think you want both. The ability to go deep when that curiosity strikes is really powerful, I think. 

[00:16:40] Alex Sarlin: That's fantastic. Yeah. And the ability to sort of allow those serendipitous moments to happen, to sort of say, Oh, you're interested in this? Well, here's 10 other things that who knows? 

And you need to be interested and allow people to have those aha exciting moments that you had in the museum. And that, you know, early readers have, right? The first book that they're like, I'm so glad I could read because I could read Harry Potter or whatever it is that changes your life, you know, absolutely does. 

[00:17:04] Ben Gomes: Yeah. I think those moments of insight are, I think, sources of great joy for people, right? And then suddenly something clicks and it comes together. Yes. And I think when you have a great teacher, when you have access to something that can give you that kind of insight. I think it makes you go a lot further. 

Yes. Because it's that joy that then propels you like, well, it's almost in some ways, something that you look for again, right? I want to be able to get that kind of insight now about something else. Now that I've learned something about rocks. What about plants? Right. I've never been interested in them, but then now I'm interested, like, how is this plant world organized? 

There's this amazing thing of like monocots and dicots that help you understand huge things about the plant world. 

[00:17:46] Alex Sarlin: Right, right, right. And you don't know the language to search, right? You don't know the terminology. So you're sort of locked out of some of the curiosity that might be really interested in. 

One thing that's so interesting about Google as a company is it's sort of, already, you know, I think somebody mentioned on stage 90 percent of YouTube, you know, people have gone to YouTube to learn something. It sort of fosters that self developed learning, that kind of motivated learner. I'm going to go search or I'm going to YouTube because I need to XYZ. 

I need to learn something, but it also potentially allows that kind of serendipitous and it allows the teacher centered learning through Google Classroom and other products like that. I know you mentioned on stage, the teacher is still a huge part of a learning experience, especially in many cases. Tell us about, you know, Those different use cases, a self directed learner, somebody's wanting to, you know, learn purely on their own, get a Google career certificate on their own versus a classroom or somebody in a, any kind of a tutor, any kind of formal learning environment and how that works. 

[00:18:42] Ben Gomes: Yeah, I think there's an interesting spectrum here. I think, and it's for many people, it's sort of a journey, I think, to go from the classroom, which is directed by someone else. And that's, I think, where the inspiration and the human plays the biggest role. But I do think humans play a role all the way through in any kind of learning, like even in these other contexts of self directed learning, the community of learners often plays an important role, right? 

Because People may think I'm fully self directed, I'm doing it purely for myself, but I'm not sure that that's fully true. In part, you're doing it because of the conversations you'll have with others, the opportunities that might open up, right? So even when you're self directed, there's a social component to how learning works in our brains that I think is something that's very important to pay attention to across this entire spectrum. 

I don't think it's something we understand fully, but I think it's something we need to make sure we're engaging with in order to make sure that real learning is happening. 

[00:19:38] Alex Sarlin: It's a fascinating point. There's a author named Martha Beck, and she had this concept of who's your everybody. So you know. It's fascinating. 

People have these self concepts that everybody will be upset if I do this, or everybody will love it if I this and then she says, Who's your everybody? And it usually is one or two people in your life, often a parent or a friend who is sort of internalized and you do everything in relationship to them. 

And when I hear you talk about that sort of internalized social. Motivation. I hear that. It's like, you know, I had a teacher who made me love X, Y, Z in fifth grade. And as a college student, I'm going to take this class because Mrs, you know, X would have loved that. And I think that's such an interesting aspect of things. 

I don't often hear about that in a technical. Concept because you think of it as person in tech, but that's really not how Google works, especially Google classroom. It's all about bringing people together together 

[00:20:24] Ben Gomes: for that learning process. I mean, I think the example is really lovely about who's here. 

Everybody. I mean, for myself, you know, we've all had teachers who inspired us for me. Partly it was my mother who encouraged us to ask questions about understanding the world around around us, but also had this a priest in. I went to a Jesuit school is based in Who I know I used to, I didn't always agree with him, but he encouraged us to always question. 

And in my head, I would argue with him, even though I've not seen him for, you know, 10 years. He was just sort of like a mental opponent who helped me clarify my own ideas. So I think that's a really important way of thinking the role of people in our lives, I don't think, can be understated. 

[00:21:04] Alex Sarlin: It's a really interesting way to sort of bridge the gap between, you know, what people may consider, and I've often considered, you know, self directed learning. 

It's like, maybe it's not really self directed. You're doing it because you were inspired by a sibling or you saw somebody on TV once who had that job and you really want to be like that. So you're going to go, I mean, who knows? There's just all of these relationships we have. So, Google is doing so many interesting things. 

You mentioned sort of that there could be a bunch of different products potentially built on top of LearnLM, which has this pedagogical content knowledge, you know, how to teach. It's been trained on, on human teachers. It's been trained on pedagogical research. What does that world look like if you were sort of to zoom forward, let's say three years, what kinds of products could come out of that powerful model? 

[00:21:46] Ben Gomes: A couple of things. I think we're seeing a lot of the evolution of our own products, right? Like in Neotear, we're seeing this ability to ask the questions that I mentioned. about the video you're watching. And can that turn into a pedagogical dialogue, right? Bringing together voice plus this sense of pedagogy, if that's what you want. 

If you just want answers, well, we have to give you the choice of doing that. But if you really want to go deeper and in that way, I think it can enable that. I think in search also, you know, you're getting through words, explanations and, and deeper versions of like, Your engagement with content on the web and guidance through content on the web, right? 

It's just, it's also rich with creative, good creators all over the world and like, and making sure that we provide you with the ability to understand and appreciate. What, what is there? So I think there are lots of use cases with classroom. We are building tools for teachers that enable them to first of all, also, you know, offload some of the jobs that are tedious for them and then enable them to do other things more creatively than they otherwise might have the time to do. 

So AI is a tool to help them there in that process. So in all of these products, existing products, there's a use of, of AI in order to. Actually make the underlying mission of that product come to life more. But I think that also new context. And I think like learn, uh, Notebook Alum is one such new context. 

And you can imagine this combining, like it's combined with voice now, but you can imagine combined with lots of other approaches to actually enrich the process of learning. A learn about is sort of purely directed at learning because that's what we were set out to do. But then again, we're looking and seeing like, okay, what is, what other things from pedagogy can we take and embed in this process? 

Yeah. How do we get to a better assessment of what is working, what is not, what's most important in this? Where we're going to get feedback from people, but we're going to have to also learn and iterate there. So I think the applications across our products and in new products a lot and then We also working with partners, you know asu and other companies to use these products in things like learn lm in their products And so who knows what they will bring what magic they'll bring to life. 

[00:23:53] Alex Sarlin: Yeah, can I double click? I want to learn a little bit more about that. So how does that structure work? So there there are people you have partners who are building Interfaces or use cases with LearnLM as the underlying model? 

[00:24:04] Ben Gomes: Well, so far we've been working with them to build LearnLM itself. Today we announced with AI Studio, you can begin to play with LearnLM as an experimental model, right? 

To see what you can do with it. But again, like I said, like we want to upstream these capabilities with Gemini. So you'll be able to do this with Gemini more broadly over time. Gemini itself is pretty, we've been upstreaming some things already with Gemini. If you could see from the metrics that already Gemini is quite good at this. 

And LearnLM is the frontier. So as we push that frontier forward, we want to get more of it into the base and core models so that everybody has access to it for the default things that they're doing. 

[00:24:39] Alex Sarlin: That's really interesting. So when you talk about upstreaming, it's sort of taking some of the core training and models of LearnLM, which is a, uh, offshoot. 

I don't know how to, the word is fine 

[00:24:48] Ben Gomes: tuning. It's a fine tuning 

[00:24:49] Alex Sarlin: of Gemini and actually pulling it back into the main Gemini Yeah. Gemini itself can be more powerful teacher. That's right. So to speak. That's the goal. And, and then maybe it can assess when people can maybe in a learning modality, right? You know, where in your, all the hours a day you spend engaged in Google products, are you in a learning, learning space? 

That's really, really interesting. I'm happy to hear that. The LearnLM is really exciting to me personally, because I feel like a lot of the pedagogical research we've known. for many years. People cite, you know, 1972 study of this and all this like chunking, you know, but it often doesn't make its way into our daily life, even in an educational context. 

Do you feel like that bringing all of what we know as well as what we will learn through AI research into sort of our daily life is one of the aspects of, you know, what are the legacies that Google might do? 

[00:25:41] Ben Gomes: I think that would be really wonderful because I think there's a lot of micro opportunities for learning in our daily lives. 

as well as, you know, you listen to a podcast somewhere, right? It's on a topic you might not otherwise have listened to, but it would be really nice if you could retain more of that and actually like incorporate it into your overall framework. Yes. And I think if we can build the tools to do that much more, you will retain and incorporate much more into your capability net framework in your mind. 

I think then you. Then, then it's possible today because I think today we are, we have access to a lot of information and consume a lot of information. Only some fraction of the information really translates into actual capability in us, right? Definitely. Yeah. And so I think increasing that pie, if you will, to be bigger is, probably a huge step forward for all that all of us can do. 

Yes. 

[00:26:33] Alex Sarlin: Yeah. It makes a ton of sense because there's so much informal learning that we do, especially as the world changes very quickly. You know, we have to read articles, we have to read newsletters, we have to listen to podcasts and watch YouTube videos just to even understand the world around us, learn new tools all the time. 

And it strikes me as like, that's one of the big differences between people who have a lot cultural capital and cultural literacy. They sort of know they have to keep teaching themselves things versus people who sometimes don't have access to formal learning opportunities. So I'd love to hear, you know, I know you think a lot about equity at Google and you mentioned sort of got your personal interest in education came from looking at schools in India. 

There's a lot of less resourcing there. What role do you see Google playing? I know there's lots of options here for sort of closing some of the equity gaps that we see in education around the world. 

[00:27:20] Ben Gomes: I think the equity gaps in education on the world are quite vast. I mean, it starts at places where there's no access to the internet and so on at all. 

You know, I was born in Tanzania and there are, there are people in Tanzania who walk hours every day just to get water, right? And so there's not time for learning. And so there's basic problems that need to be solved various places. So, but then there are, there are a lot of people who don't learn to read. 

There's a lot of people who go through school and just never learned to read and they don't learn basic math. We have this protocol read along where already about we've working with nonprofits again. With other people involved, we've shown that about a million students have increased their reading level in India. 

And so, and this is an AI, an AI product before the current wave, it was built with speech recognition to help this somebody read out a text and then point out. where there might be errors, right? And so it was built to also be on device, so it could be done in a cheap way, on phones that are available without in head connectivity, and so on, right? 

So that is sort of, uh, that took years of work to build that, and I think we will be going through those paths from where we stand today to first of all building out the effective ways of learning, and then making them be broadly available. So it's going to take time and steps, but I find that prospect really exciting because I can see over time, smaller moms are becoming more powerful. 

The capabilities are increasing. And so you're going to be able to do much more with much less. And so the, I think there's a huge potential for us to unlock these capabilities, not just at one end of the spectrum where people have access to everything today, but at all levels. For the whole spectrum, and I really hope it will become much more of a level playing field in many places in the world. 

[00:29:07] Alex Sarlin: Yeah, and you know, Google Android played such a essential role in bringing the cell phone revolution to the world. It's sort of moving it away from the high priced, you know, Apple model. It feels like those small language models on device could be transformative. It could be an operating system. It could be in apps. 

It could be tailor made. But. That's an exciting future to think about. 

[00:29:28] Ben Gomes: It really is. Right. I mean, I, when I was working on search, I joined Google in 1999. And, but in like, I think this year or so later, and, you know, people over here had not really heard as much about Google even, but then I was on a bus between, in two, between two cities in India. 

And people asked, somebody said, you know, I've been using Google. I was like, That's amazing. This thing that I thought was this project being worked on somewhere in Mountain View is actually finding use for people in the other, on the other side of the world. And like that potential for technology to actually help people in many different places, I think is really quite amazing. 

Very motivating. It's like wonderful. 

[00:30:05] Alex Sarlin: Yes. And I think it unlocks that sort of proactive model, right? If you have a phone and you're in and you don't have to necessarily go searching for things, right? It can say, Hey, you're, you're in this place. Do you want to learn about this place? Or, Hey, you seem to be doing this kind of work. 

Do you want to learn more about this kind of, I mean, there's so many interesting places to sort of. Engage people in learning proactively with AI, which is just thrilling because I think it just opens up doors everywhere. 

[00:30:31] Ben Gomes: I fully agree with that. I think that we have to find ways to make it make sure we keep it engaging. 

Yes. So that people actually like us for it because and so. Keep it engaging and every experience that they have that they will, they find engaging will cause them to do more of it. That's what happens when learning actually works. You want more of it. Completely. And so I think we have that potential now to start creating those experiences. 

[00:30:54] Alex Sarlin: Completely. This is a very silly metaphor, but I remember, you know, my first ride share was, I think Uber at the time was giving out free rides for a weekend. And I was like, Why would this is so silly, but of course, if it's a free taxi ride, I'll take it. But they knew that having that experience, that sort of whole loop of, I call the cab, it showed up in one minute and I got there. 

Oh my gosh. Would unlock everything. And I feel like that sort of first experience and it just working. I mean, Google's had that in so many different arenas with, I remember when Gmail was in beta for eight years, but it was like, everybody went to Gmail because it worked, it was free. And it worked incredibly well. 

And you had space. And, you know, I think Google has an incredible track record of sort of shipping products that change the way people live. The potential is sort of endless when it comes to AI and learning, especially. 

[00:31:40] Ben Gomes: Yeah, I think the potential is really enormous. I think associated with that thing of like the first experience also feeling that it worked for me and then it was good, right? 

Like I think a lot of students struggle in any subject when they find like, Hey, I tried that. And it didn't work, right? And then the next time they're more hesitant to try it. And so I think finding that unlock, so it feels easier. It may never feel easy, but if it feels easier and that bit of challenge to be overcome is overcome, they can overcome it. 

I think there's. There's huge potential there for them to go off to the next challenge, the next challenge, the next challenge. 

[00:32:17] Alex Sarlin: Yeah. This is self efficacy theory, right? You have a mastery experience, you feel much more excited and motivated and more confident to try it again. Or a vicarious one. You see somebody else doing it, the same thing. 

And so that there's a virality there. But this is so interesting. Yeah. Interesting. This is Ben Gomes, the Senior Vice President of Learning and Sustainability at Google doing incredible work for the world's learning ecosystem. Thank you for being here with us on EdTech Insiders. 

[00:32:43] Ben Gomes: Thank you so [00:32:44] Alex Sarlin: much for having me. 

In this next conversation, we're speaking with Maureen Haymans, the VP of Engineering in the Learning and Sustainability Organization at Google. Maureen leads initiatives to revolutionize learning and education within Google products such as Search, YouTube, Classroom, and more. Her work centers on harnessing AI to create personalized tutors for every student and teaching assistants to empower every teacher. 

Ultimately, democratizing access to quality education, so anyone can learn anything. Additionally, she is a lead in Google's sustainability organization, harnessing information and AI innovation to identify solutions for combating climate change and helping build a sustainable future for everyone. 

Previously, she led the Google app and discovery engineering team and worked on improving Google ranking for many years. She started in Google in 2003 after getting her master's degree in computer science at the University of California, Santa Barbara. Maureen is originally from Belgium and moved to the United States in 2001. 

She completed her master's degree in engineering with a specialization in applied mathematics and a minor in computer science, at the Université Catholique de Louvain. We're here with Maureen Heymans. She is the VP of Learning and Sustainability and does a lot of engineering work to make all of Google's amazing work available to people around the world. 

Welcome to the podcast.
[00:34:11] Maureen Heymans: Thank you. Thank you for having me. 

[00:34:13] Alex Sarlin: So, our first question is, Tell us a little bit about their learning and sustainability team and how you got into the sort of education world. How did you get into education in Google? 

[00:34:24] Maureen Heymans: Yeah, so I have been at Google for almost 22 years, you know, pretty much came fresh out of college. 

I always joke that I had to apply twice. Google was my dream job and I didn't give up. It didn't get the job the first time, but yeah. And I started working on search, which was kind of my dream, you know, the core mission of Google and through the years I work on improving our core ranking to building the Google feed. 

Um, Which is this personalized feed of content. And the mission of the Google feed was actually to really connect you with engaging content so that you can nurture your curiosity and really discover new passions. And so it was already pretty close to learning and curiosity. And then during the pandemic, you know, as many parents, I got more exposed to my kid's education. 

We also decided to pull them from public school, you know, not much to lose and decided to put them into this like startup, like experimental project called Prisma that was all about project based learning. And that was something I was always interested in. You know, I'm a strong believer that if you can connect learning to real world problems and do interleaved learning, it makes, you know, learning so much more engaging. 

And so we did this experiment for six months and the kids were. So motivated. It was amazing to see how much they wanted to learn. They made a project like, you know, citizen of the future where they had to learn about climate change, but then also learn how to build like disaster proof building. And so they had to learn about all the disaster in California, many of them, but also use like tools to build and then present the finding. 

And anyway, all of this really made me super excited about learning. Um, And at that time, actually, Ben Gomes, who used to be the SVP of Search, left Search to start a team just on learning education. And so I decided to join the organization three years ago. And it's been super exciting. You know, I still get to work a lot on Search, improving Search for when students come for the learning needs, but also getting to partner with YouTube on improving learning on YouTube even more active learning, but also Gemini, you know, one of the newest product, but where we see so much adoption from learners and then partnering with many, many teams across Google on how we can improve learning. 

[00:36:47] Ben Kornell: Yeah. I'd love to go deeper on that. So the learn X team, which, you know, Google is somewhat famous for how complex the organization is and different verticals and the challenge of different teams talking to other teams. You have an explicit mandate of actually working cross functionally across the whole org, uniting, learning, and thinking about it holistically. 

What have you learned from that process? How's it changed the way the Google team is approaching learning, especially with new AI tools? 

[00:37:16] Maureen Heymans: Yeah. And I will say LearnX when we came up with the name, it was both like for 10X innovation, but also cross teams partnership. So, yeah, I mean, first, if you think about learning, it's, it's really called to Google mission of organizing the world's information and make it university accessible. 

Right. So no surprise that many teams across Google work on learning. It goes from learning for school, like classroom, but also many students coming to a product. To learning for work with like a project like grow with Google, we provide credentials, but also cloud learning where you can acquire a new skill set for your job, but also learning for life, right? 

So many people come because they want to learn a new topic. They want to learn a new hobby. And of course, YouTube is an amazing platform for this. And so when I joined the org, I knew it would be an interesting effort where you have to work with so many teams across Google to Try to do more together, right? 

To try to come up with shared principles, to come up with shared priorities, figure out how we can share technologies and insights. And so one of the things we did and big credits to Julia and the pedagogy team was to come up with learning principles. So we had lucky to have a team that has lots of pedagogy expertise and they taught it a lot because a lot of us are new to, to learning. 

But really came up with those principles that really we want to infuse across all of our product at Google. We want to be bold and leveraging the technology, but we also want to do it in a responsible way. And so, you know, with those learning science principle, it really makes sure that we're not just We are optimizing for learning outcomes and making sure that, you know, what we are building makes sense for students. 

And so some of those learning principle includes, you know, managing cognitive loads, making sure that when we provide information, it's well structured. It's kind of break down problems and includes making sure we 

provide active learning. You know, as I mentioned, we want to make the experience more interactive with conversation, with practice problems. 

It also includes encouraging metacognition, right? We don't want to do too much on your behalf, like we, we want to provide learner agency so you can come up with your own strategy and really think of how you are learning. It includes curiosity and motivation. Anyway, I don't remember all the principles, but those are the type of thing we can do as a central team to make sure that as you know, teams across Google build those learning product, they really take that in mind. 

So that's one way we are contributing as a central team. Another way is actually through a partnership with research in DeepMind and through technology. So, um, you know, we are lucky to partner with some of the experts in the AI domain. But we want to make sure that when they build those foundational models, they work great for learning and they really infuse those pedagogical principles. 

So we work with them by providing them the data so that they can train the model. We also provide them with evals, you know, STEM reasoning is a big one that we want to make sure that if we are going to help students on the math problems, we actually do it in accurate ways. So we partner with them to make sure they can improve that. 

But we also want to make sure they can do great on like text leveling or providing practice problems that make sense, uh, or being able to do grounding. And so we really partner with research to make sure that our foundational model are working great for learning so that we can leverage them in a product in a shareable manner. 

But we can also make it available to third party partners to build great learning products. 

[00:40:47] Alex Sarlin: Yeah, you mentioned interleaving just in passing. And then what are your answers? And I was like, wow, that's a great natural pedagogical principle. It's obviously sunk its way into your whole team understands sort of the pedagogy of this stuff. 

So let's dive into the last part of your answer there. You say, make it available for others who want to build on some of the amazing work that Google has done. There was just an announcement at this summit about LearnLM, your fine tuned learning model. Having an API attached to it, the developers can use to actually build on top of. 

Can you tell us a little bit about what that looks like and how EdTech entrepreneurs or EdTech developers, many of whom listen to this podcast might start using it and trying it out? 

[00:41:23] Maureen Heymans: Yeah, as I just mentioned, right? I mean, we have been working closely with research and deep mind to make sure our models are performing great for learning capabilities and really infused pedagogical principles. 

And so. We have been testing those models internally, but now we know that, you know, Google is not going to solve the full education problem on its own. And so we want to enable many partners outside of Google. And so. We are taking a first step of doing that by making those models called LearnLM available through AI Studio. 

So AI Studio is a way for a developer to access some of those technologies from Google and quickly experiment, see how it works. And what we hope to do with this pilot is really get feedback, right? We want to get feedback on how The models are performing what we can improve. Maybe we'll discover new use case that, you know, entrepreneur exploring. 

And so, yeah, super excited about this first step of making it available to more people. [00:42:19] Alex Sarlin: Yeah, it's incredibly exciting. 

[00:42:20] Ben Kornell: Yeah. I'm so curious, stepping back, you know, you talk a lot about learning and it's called the LearnX team and this conference is all about learning and when we talked You know, one of the things the team debated was the difference between education and learning and this sense that Google essentially can make learning anytime, anywhere possible. 

How do you think about the interaction between learning and education and education systems and how does that come into play in how you build your products, tools and systems for those systems? 

[00:42:54] Maureen Heymans: Yeah, that's a great question. So, of course, you know, we think teachers are central to the learning process, right? 

And so that's why we, as we think about learning, we really make sure that those tools are vetted by teachers. And we really empower teachers with the right tools. And so that's where we are partnering closely with the classroom team on providing a lot of those tools. in the classroom, both like tools that can make, improve teacher productivity, lesson plan, you know, practice, but also learner tools that they can decide that they want to enable in the classroom. 

At the same time, we know that Students are in the classroom for a few hours a day, and then they still need to do a lot of learning outside the classroom. And a lot of time, what they do is they go to some Google product to get that support, right? And whether this is for the homework, you know, we see a lot of homework queries on search and Gemini, but also because they couldn't understand a concept, and so they go to YouTube. 

And so we feel like that's a continuation of the learning. And that's where we can support them 24 seven. And they have that tutor that can help them, that can be completed, judgment free, and can really over time adapt to the learning needs, right. And And so that's why we've been super excited to, to work on improving the learning experience on search and YouTube and Gemini. 

So we can extend that learning outside the classroom, but, you know, of course, trying to connect it back when we can with the classroom. So, so something that for example, we are looking into is we would love to adapt learning based on the student's textbook, right? Or curriculum that. The teacher provided. 

So imagine that a student could, you know, either say that's my textbook and we have partnership with the publisher, but also I'd take picture of the curriculum or the notes from the classroom or, or like whatever content that the teacher provided. And now we can automatically generate some study plan to help them get prepared for a test. 

Or we can generate practice problems based on what the study in the classroom. And so you can imagine that. You know, for those students that, you know, might not have understood everything in the classroom, or need to continue practicing, we can continue that learning journey outside the classroom. So I think that that's super exciting, you know, and if we can connect the two even more strongly, I think that makes it even more powerful. 

[00:45:11] Alex Sarlin: Yeah, it's a great point. Students are sort of already bridging the gap between Formal education and informal learning by 

[00:45:17] Ben Kornell: Googling. Yeah, it's a pre existing user behavior. So you're really enhancing something that students are already doing. 

[00:45:23] Alex Sarlin: Yeah, yeah. The YouTube folks said that, you know, a huge percentage of college students come back after their first day of class and use YouTube to supplement their learning and learn what they're good, you know, sort of make sense of it. 

So the idea of making that easier to connect and accurate and make sure it actually ties into, I mean, it's always accurate, but you know, ties into what they're actually learning in the context they're learning. 

[00:45:41] Maureen Heymans: And I think, especially if we can ground that on how do we, they learn in the classroom, that's even more powerful because sometimes it's multiple ways to explain the same concept or like based on your district or the country you live in, it's going to be explained in different ways. 

You know, the way my kids are learning, I read medics is completely different to the way I learned it. So, but now if we can ground that. To the way that they have learned it in the classroom would make it easier for them to build up on that, uh, prior knowledge. 

[00:46:07] Alex Sarlin: You've mentioned project-based learning and one tool that Google has put out that I think it's flown a little under the radar, but it's really fascinating, is this learn about tool. 

[00:46:16] Maureen Heymans: Yes.
[00:46:16] Alex Sarlin: It sort of bridges search and learning and it uses learn. Lm can you tell us about that 

tool and, and you know, sort of the origin of it and what it does? 

[00:46:24] Maureen Heymans: Yeah, that's a tool I'm super excited about. I'm proud to see my team build it. So super cool. But really, the way we look at this was that, okay, if you look at search, you know, we have the power of the web, right? 

The web is so rich. There's so much great content. How can we pair that with the power of AI? And then also infuse learning principles. And so that's how we came up with Learn About. Looking at how people learn the best, right? And so as part of this experience, you can have a conversation with a tutor. And we definitely see through our research that that's a big part of why students are so attracted to AI tools. 

They want to be able to ask follow up questions and to have a conversation. So this is a conversational experience. But we also try to provide you with a bunch of tools that can help with learning. So first we have 

this lesson plan on the left, which is kind of a scaffolding. So you can get a sense of the space, you know, you can get a guided conversation, but then we also, you know, in the experience bring those what we call learning resources. 

So it will have things like You know, interactive lists, which kind of break down problems into subconcepts. It will bring things like why it matters, which I think for me is so critical to explain why this concept we had learning in school, you know, matters in the real world. It will bring misconception. 

It will bring vocabulary. And it even brings what we call lightweight practice. So that little quiz to make it fun because everybody likes to, to do a quiz, right? And so, yeah, by bringing all those learning principles in the product and then bringing the best of the web and the best of AI that brings conversation, you know, the team has managed to build this learn about experience and it's been pretty amazing to see the positive reaction, like reactions, both internally and externally and how people are using it, both for curiosity, but also for learning. 

And, and so they feel excited. to bring it to more users. It's available on Google labs, but we do if you want to make it access more accessible. 

[00:48:17] Alex Sarlin: Yeah. And it sort of starts out looking almost like a search experience. You say, what do we want to learn about? And then you realize, wait, no, this is a conversation and it's generating all of these really, you know, guided specific tools and resources to help you learn. 

Really fascinating. I mean, we're in the field and. Didn't come across it until relatively recently. I don't know how long it's been out. 

[00:48:35] Maureen Heymans: Yeah. I mean, that was also a fun project because we decided in this world of fast evolving technology, we really need to embrace this culture of rapid prototyping in the team. 

And we're like, you know, things are moving so fast that if you have a six months plan, by the time you're done, it's obsolete. So we really tried to do a rapid sprint where we kind of every week try it out. It's some, you know, UXR and to get feedback. And. And really iterate quickly. So the team was able to do this, build this. 

And like, I would say like, get a quick demo in like two or three months and then iterate and get it out on, on labs in a little bit less time and a little bit extra months. And, and so, yeah, so it was really fascinating how quickly they were able to iterate. One thing I will mention that is super cool is actually. 

Most of this was built with just prompting and symbol technique, like rag. And so you didn't, we didn't need, of course, build on amazing models that I got from Google, so, but the team didn't need to do like expensive fine tuning and, you know, we just leveraged the underlying model prompting, you know, the fact that the models are good at instruction following and. 

And really play with that big prompt. And so now, you know, it's, it's fun to see that those code, that's just a big prompt, right? And, and so that's also enabled us to move super fast. 

[00:49:50] Alex Sarlin: That's amazing. 

[00:49:51] Ben Kornell: Yeah, it's incredibly compelling. And it's also just an example of how much Google has changed operationally in their, the ability to ship products and be iterative and be nimble. 

And I think LearnX is actually one of the best examples where. Working cross functionally, moving fast, like, providing value. You know, I think there's an optimism at this conference, because of what's happened over the past year or two, about what could happen in the future. As you look forward, you know, five, six, seven years, what does success look like for LearnX? 

Do you think the world will look like for learners? 

[00:50:26] Maureen Heymans: Yeah, it's a great question. And one thing that actually reminds me is one that I wanted to mention when we were talking about Linux is I always say this team has like some superpowers that make them successful. You know, one of them is flexibility and agility of working across so many different teams, but also collaboration, right? 

And innovation, trying to see what's possible in six months and not what's possible today. And so that's really also how I kind of see the success of this team. It's not just what we are going to do. On our own as a little team, but it's how much we can inspire other team at Google to do more in the education space, right? 

Because Google, like you always say, is one of the biggest tech company in the world. And so if you can, Enable more people at Google to build those great learning experience, you know, that can have a huge impact, learning impact across the world. So that's really how I see the role of my team is to enable other team to build together, to, to get great foundational models so we can make it available externally as well. 

That, that's really the success. No, more broadly, I guess that's your other question of how AI can impact learning, right? And I'm definitely really excited about how we can build personalized and adaptive learning experience going forward, right? That's always the dream of a teacher, but really hard to do because you have to support so many different students, right? 

And Now, by looking at the context, right, the learning context, like what curriculum they use them, you know, what's, what's the misconception, all of that, you know, looking at the interest, looking at the goals, you know, we can really provide an experience that is at the right level. Kind of offer productive struggle, connect them to things that they care about. 

And we know if students care about something, they will do much better, you know, in face of challenge, right? They will care more, they will put more energy into it. And so that's the part that I'm super excited about learning is that we are really going to be able to build this personalized and adaptive tutor for every learner and, and. 

Do that in a equitable way across the world because there's more people have access to technology and thanks to the phone and Wi Fi, you just need the phone right to have access to those tools. And so not everybody has the chance like in the Bay Area to have access to amazing teachers and and so now if you can provide that access across the world, that can have a big impact. 

[00:52:45] Alex Sarlin: And there's such a range of places to do it. Google Classroom. We've got to go with it, but Google Classroom, you know, I could imagine one touch differentiation in Google Classroom, a place 

where you already have a class and individual students that you know about them, that you might know their interests, or you can do it on Android, where you have people all over the world. 

Exactly. It's just such an exciting. Yeah. 

[00:53:05] Maureen Heymans: And, you know, I always feel like as a parent, my kids are maybe not the best. Use up those credits because I'm there, they have a great teacher, I can help them. And then of course, there's lots of curiosity, we can do more. But when you look at, you know, students across the world, in India, where, you know, you don't even know if the teacher is going to show up to the classroom, you know, that's where, I mean, of course, human connection is so important, but when there are no humans, that's where, you know, those AI tools can, can be so beneficial. 

[00:53:30] Ben Kornell: I think there's a day where billions of people have access to high quality education that have never had access to it before. And I know that Google is probably not going to be all of the solution, but the work that you're doing at LearnX. provides such a great baseline for others to build and to empower the learners to power the educators. 

It's been so fascinating to watch. Maureen Heymans, VP of Learning and Sustainability. Thank you so much for joining us today. 

[00:54:00] Maureen Heymans: Thank you. It was so fun talking to you.

[00:54:03] Alex Sarlin: Thanks for listening to this episode of EdTech Insiders. If you like the podcast, remember to rate it and share it with others in the EdTech community. For those who want even more EdTech Insiders. Subscribe to the free EdTech Insiders newsletter on Substack.