The Evolution of Google for Education with Shantanu Sinha

Shantanu Sinha is the VP and GM of Google for Education. In addition to bringing Google products like Workspace to schools, Shantanu and his team build educational products such as Google Classroom, which helps over 150 million students, teachers and education leaders across classrooms around the world, and new experiences such as Practice Sets and Read Along, that demonstrate the power of artificial intelligence for teaching and learning.

Before Google, Shantanu was the founding President and Chief Operating Officer at Khan Academy, where he teamed up with college roommate Sal Khan to build an organization that helped make personalized learning globally accessible and free. Shantanu is also a founding board member of Khan Lab School, where all three of his children attend school.

Shantanu has leveraged his strategy and operations expertise from his time at McKinsey & Company, along with his background in computer science, math, and brain cognitive sciences from MIT, to make an impact in an area he is most passionate about: education. Follow @ShantanuKSinha.

Recommended Resources:
📚 Google Keyword blog
🌐 Google for Education website
🔗 @Shantanu Sinha on Linkedin, @ShantanuKSinha on X

Transcript

Ben Kornell: All right, EdTech insiders. We have a very special guest. Someone I've known for several years now who has an amazing education and impact experience and trajectory and career who's now joining us, Shantanu Sinha. Welcome to EdTech insiders. 

Shantanu Sinha: Great to be here. Thanks for having me. 

Ben Kornell: Before we go too far into what you're doing today, actually could you tell our listeners a little bit about your professional path?

How did you end up leading the team at Google for Education? 

Shantanu Sinha: Yeah, I think I've long had an interest in education when I was in high school, used to tutor when I was in college, used to volunteer at the local elementary school, but really what brought me to the space of education was when I helped get Khan Academy started as a not for profit organization back in 2010.

At the time I had known Sal for a really long time. We were friends in high school. We actually first met in a high school math competition down in New Orleans, Louisiana. We were roommates in college and worked in some ed tech, worked in some tech companies in Silicon Valley, the late late nineties and gone off in our Parallel pathways.

And he was at a hedge fund and I was at McKinsey and company. And he had, I won't go through the full origin story of Khan Academy. Many people have heard that, but he basically had this YouTube channel that was taking off. And we had this vision that we could create a new type of organization.

And I joined to help really build the Khan Academy as an organization with this audacious mission of. Providing a free world class education for anyone anywhere, and we felt at the time that technology was really transforming what was possible, and it would be really valuable if we could create a not for profit that would just totally focus on making sure that educational materials are as available to as many people as possible around the world.

Ben Kornell: So tens of millions of kids weren't enough for you. And then you're like, where can I go where it's even bigger and bigger reach? Actually, I'm curious what prompted your switch over to Google. And tell us a little bit about your journey at Google. 

Shantanu Sinha: Yeah. And exactly. I think the ride at Khan Academy was quite remarkable.

And that's what really brought me into the education space. And then I did that for five or six years and then joined Google about eight years ago. And what drew me to Google was really the scale and the technology that an amazing platform that provided so much of how people learn really starts with search or YouTube and, many of our products are integral to the learning and education experience.

Chromebooks classroom workspace I thought it was just a remarkable opportunity to take these huge, massive products with incredible reach and figure out how we can improve the teaching and learning experience through them. And yeah, I was also really drawn towards the idea of working with the technology, the cutting edge technology that a place like Google could provide.

I was always really interested in artificial intelligence. I should do my master's thesis in artificial intelligence at MIT many decades ago. And the idea that Can work in an organization that is at the cutting edge of technology and has a platform for scale. I thought, there's very few places where you could have that type of impact in the world.

And that's really what brought me to Google. 

Alex Sarlin: Yeah, you mentioned Chromebooks and Google Classroom and Google Workspace. These are some of the most ubiquitous tools in all, everywhere. And including in education, the newest pieces is Gemini, Google's most sophisticated LLM. So you mentioned artificial intelligence.

Where do you think we are all headed with Gemini and AI over the next few years? What makes you excited and what makes you concerned? 

Shantanu Sinha: I'm really excited that the advances and the innovation that's been happening in the field, as somebody who's been an AI enthusiast for quite some time, it is just absolutely remarkable to see the speed of these advances.

And I spent a lot of time playing with Gemini and really excited at how we're pushing forward. The underlying science and in many ways Gemini is multimodal by its very nature. It was built that way. It has new capabilities like large context windows where you can now provide a. million token prompt into Gemini and even larger in, in the research lab.

So it's really unclear where the limit is on this and I think that opens up totally new possibilities. You can just imagine rather than asking when you're chatting with a chatbot for it to from pre training try to predict the responses, if you can say why don't you read these Books first, and then give me an answer.

The same way I would probably answer a physics problem a lot better if I had a textbook I can read before I answered it. The same thing is true with these models, and I think that opens up all kinds of different possibilities, and I think it's really early. I still believe that we are for it. Very much at the early phase of what's going to be possible with a I, people like to say these are the worst models we'll ever have in our lifetimes that we see today.

And it's really exciting to see how the field evolving what it's opening up. In terms of areas that I'm, more worried or cautious on, I think one of the biggest things with technology that's moving this fast is really understanding what's possible. What are the capabilities?

And how ready it is. I've never been in a world where things went from the research lab to scale deployment And sometimes in a matter of months. That's really an unheard paradigm. And I think when you're dealing with something like education, which is fairly high stakes for students, for so many people, it's so important to get that right.

So we like to say we want to be very bold and responsible in how we bring this technology to our user base. It's really important that when people are building products, that they really understand the underlying capabilities of the technology And they're orienting their use cases on what the technology is good at and being very transparent with users on what it's not good at and making sure that they're optimizing those use cases appropriately.

For example, we all know that many of these models can hallucinate and they can create that kind of information that may not be that factual, in some places if you apply it that can be a feature So if you make it as a creativity tool and this can really help you Brainstorm and think of new ideas that suddenly that's a great way to apply it But there are a lot of places you can apply it where you know There could be a real cost to getting things wrong And I think when you're thinking about bringing things into education It's so important to understand the technology, understand the use case and make sure you're deploying it in a way that is as responsible as possible.

Ben Kornell: As you mentioned, there's been this drumbeat of from R and D to beta launch to global launch. That's been pretty unprecedented, and it feels every 3 to 6 months, Google has actually been a part of that new release. Are there any new announcements on the horizon for you, for Google, for the team in the generative AI space, especially with regards to education?

Shantanu Sinha: Yeah, so quite a few things that we're really excited to bring. So we are bringing Gemini to our education customers. And there's three big parts of that we'll be bringing the Gemini, Gemini education add on. So as a part of Workspace. So if it's a Google workspace for education customer, they can purchase this add-on and really infuse the power of Gemini throughout Workspace.

So things like, summarizing emails or helping a teacher write emails, creating lesson plans and docs. Creating a slide presentation for your next class, really bringing the full power of generative AI into the broad capabilities within the workspace portfolio. The second thing that we're excited is to announce is that we're bringing the ability for admins to turn on Gemini at gemini. google. com so that their user base can chat with Gemini free of charge in a data protected way. So as we all know, data protection is really important in education. So when users are chatting with an AI chatbot it's really important that you understand what's, how that data is being used, what's going on? What are you? What are you? Where is this transcript going?

So with our new advanced data protections, our education admins can be very confident that data is not being used for personalized advertising. It's not being. It's never reviewed by humans. It's not being used to train the models and really is done in a protected way. And the third thing we're actually bringing a number of education, learning and education focused enhancements.

to Gemini a suite of improvements to the experience. One of the things I'm excited about is new partnerships we have with data commons and open stacks, where when you are chatting with Gemini, you can actually ground your responses in data, like hard data from data commons or in open stacks textbooks.

So you can really be sure that your interaction is more factual in nature, but also if you're using an OpenStacks textbook in your school, it's as relevant and appropriate for your class as possible. 

Alex Sarlin: Those are incredibly exciting announcements. I'm particularly partial to the OpenStacks partnership.

That's like a huge leader in the OER movement for years out of Rice University. They're an incredible group, and I'm so happy to hear that the OER space and Google AI suite are really coming together in some of these products. That's amazing. So how do you see AI integrating across the Google for education suite in the future?

We often talk on this podcast about how Google has this incredible advantage and that everybody's already using so many of your platforms and systems. And as you just mentioned, the AI can enhance them further. You don't have to convince somebody to go somewhere else to try it. Do you, is that going to continue to happen?

And I'm sure the educators on the podcast are curious, listening to the podcast are curious. Will there continue to be free versions of the Google for education products and features in the AI era? 

Shantanu Sinha: Yeah, I really believe that the AI is this transformative type of technology that has an unbelievable potential to elevate the educator, really make them much more productive, make them much more effective and efficient in their job.

It's really important for us as we deal with this new technology that it's, we understand how it can be applied. In the workflows of our users, and I really focus on the users. In this case, the technology you never want to build for the technology sakes. You really want to build for real user needs and real challenges for your end customers and your end users.

And we're going to continue to find ways that we enhance productivity. We enhance the teaching and learning experience in the classroom. We enhance how people can access this through chromebooks and everything across Our broad portfolio of products that we have at google including search and youtube we really do see that it's important to Give people options.

So We're going to always have a free offering of workspace and we invest heavily in that and we think it's a really strong offering for schools and institutions. And like I mentioned, we're enabling the Gemini experience. People can chat with Gemini. As a part of that free offering so that really, people can start to experiment in the space and find ways to find journeys that really work for that.

And then we're going to continue to make sure that we can bring the best of AI to our users. And in that case, it is a broad set of functionality across Chromebooks across workspace across classrooms. We think it's important that we enable our customers with the best that this technology.

Can bring, but we're also focused on making it as accessible as possible to as many users as possible. 

Ben Kornell: And Shantanu, when you've given examples of this, a lot of what you're talking about are the administrator or educator users. Do you foresee kids being able to have access to Gemini? In the Google for education suite as well.

Shantanu Sinha: Yeah, so we're it's an area. We're definitely actively looking at our initial rollouts are focused more on 18 plus. So higher ed or administrators, it goes back into the point of really trying to make sure that we. Bring this as responsibly as possibly to our user base. And in many cases teachers want to try this technology first.

They really want to understand it. They want to be able to use it. They want to have a really strong sense of the capabilities before enabling that for their students. But we're actively also looking at what we can bring to younger Users as well and making sure that whatever we do bring is has the right controls for teachers so that they can, they, they can embed it in their classroom in a way that works for them that really is putting them in the journey, putting them in the loop in the right way and giving them the right management controls and it's done in a way that is as safe as possible for students.

Ben Kornell: Yeah, this really builds on your announcement at bet where you launched a whole suite of AI enablements, mainly around engaging and interactivity in terms of content. And just in some ways it seems like a huge leap forward from where Google was just even a year ago, but then it's we're in this new era where every three months there's just new rollouts.

I would love to just, take off your Google hat for a second. Just step back, looking at the space more broadly, as you're looking at ed tech and the education and technology landscape, what is the most exciting trend on the horizon that all of our listeners should keep an eye on?

Shantanu Sinha: Yeah. So I think that obviously there's so many different ways that you can apply generative AI and AI to these different journeys. One of the things that I'm most excited about is the unexpected applications of AI where you can, Enhance creativity in new ways. Create more interesting, more enjoyable, fun, novel experiences in classrooms and for users on.

I'd actually really encourage people to when you're dealing with technology that is this transformative. It opens up a whole different space of thinking. I've definitely seen it with my own kids where you go in and you say Okay. Maybe you're going to use generative AI to help or use one of these large language models to get help on your homework assignment.

But instead they're sitting there and they're writing new stories and they're creating videos and they're creating projects. And they're just like interacting with this technology in a totally new way. And I've just been amazed to watch. My own kids through that, where whether they're programming or creating videos or write or, the creative writing is, it can end so much.

And I do think that's an area where really exploring the new teaching and learning interactions that can happen when you and how you can rethink that. Make much richer experiences is something that I'm really excited that people are experimenting with that just around the around the globe.

And I think that has all kinds of potential to really make the whole educational experience so much more engaging, so much more interactive and so much more enjoyable for students. 

Alex Sarlin: It's a really great point. I mean that we always talk about the Bloom’s taxonomy and education and ed tech and the top layer of that is creation.

And I think maybe it's that I love hearing you talk about. It's underappreciated how much a generative AI basically turns out on its head. Creation is the first thing you do. You make up new stories, you make new images, videos multimodal. And I think Google is really at the forefront of that.

I think it's going to change how education is. is perceived. You can create before you even know what are the underlying facts and foundations. It's really quite amazing. What is a resource? We have to ask, we've talked about a lot of things in this interview already. Privacy. We've talked about the Google suite for education, which is really widespread.

The reach, your partnership with Khan Academy and Khan, which is amazing. What is a resource that you would recommend for the listeners of this podcast who want to dive deeper into any of the topics that we already discussed today? 

Shantanu Sinha: Yeah. So a lot of the announcements that we just talked about today we can find more on our keyword blog.

And we also have really robust Google for education website with a lot of materials on all of the different resources of what we bring to schools and institutions as well as all of our latest announcements. You can also follow me on LinkedIn and X as well as Google for education on, on, many of the different social platforms.

Ben Kornell: Awesome. Shantanu, such a pleasure to talk to you as always. Really, it's amazing to think about, how 14, 15 years ago, you were really starting Khan Academy and here we are in the AI revolution and you're leading the charge as GM at Google for education. It's just so inspiring. Thanks so much for sharing your insights today and joining us on edtech insiders.

Shantanu Sinha: Thanks for having me. Great chatting with you.

Ben Kornell: Now we've got our bonus after hours with Shantanu. This is our member, for members only, but we're going to give a sneak preview to everyone. So everybody gets a chance to hear some follow up questions. Shantanu, you just dropped some big news around the integration of Gemini. Into the Google suite. And so Alex and I just wanted to ask some follow ups first, let's talk a little bit about concrete classroom examples.

I'm the average teacher. What does this mean for me now that I essentially have Google's AI model at my fingertips and I'm using the Google suite. 

Shantanu Sinha: Yeah, so the first thing is there are so many different journeys that you can do every day as a teacher that you can start to experiment and say can I help me with this?

If I'm lesson planning for my next day's class, if I'm trying to create a good project for my students, if I'm trying to grade a set of Homework assignments. If I'm really like just looking for a nice hook to engage my class. All of these are things that you can now go to Gemini and ask for help with or in the context of our Gemini add on.

If you're using workspace, just embedded in those experiences while you're in the process of making your slide deck, you can start to have that conversation with AI and really start to enhance your materials. 

Ben Kornell: So I'm like in Google Docs or I'm in slides and I can generatively make that next unit. How about Google Classroom?

Is, am I going to be able to use generative AI on one of my tasks or cards in Google Classroom? 

Shantanu Sinha: Yeah, so one of the things that we've enabled is a new labs. Tab. So we have a pilot program for people really want to test out the newest in teaching and learning that they can sign up for. And as a part of that, there's a new labs place in classroom.

Where we have new experiments that we're running and new ways that you can try out some of these journeys. And it includes things like lesson planning. It includes things like being able to take content and re level it to your class. So if you have material that you have from a different source, or maybe you found online, but you want to make sure this is really relevant for my third grade class.

You can use generative AI to rewrite that. So we have a number of things that are really focused on teaching and learning journeys and experimenting with that with the classroom. And then if you are using our products, Within Docs, Slidesheets, Gmail, you can also bring in these experiences there as well as just chatting with Gemini directly at gemini.google. com. 

Alex Sarlin: I wanted to double click on what you said about the context window being so enormous and the OpenStacks Data Commons partnership you're thinking about. You said this very quickly, but it's a very transformative concept in generative AI, the fact that you can train a model, fine tune a model in real time on any particular Content a whole textbook at a time.

Maybe you can unpack for people who don't know the token level. What is a million tokens? How much can you put into Gemini before asking for an answer? And what do you think teachers are going to do with that power? 

Shantanu Sinha: Yeah, so the tokens is a little bit of a weird word, but you can think of it as you could have a word, but also a period or different things.

These models are really just trying to predict the next token in their responses. So when you write a prompt to one of these models there's only so much information that you can predict. Provide. So sometimes you might just say, Hey, help me or help explain this topic. But when you have the ability to have these really large context windows, it is game changing in that way, because historically, the only way you could Is you have to run this fine tuning process where you have to fine tune these models of all this new data, but now you can really just pass that into the context window and it makes it so much more accessible.

So I'm really excited about what's possible with that. I think you're going to see a lot of ed tech providers. You're going to see a lot of teachers, people in the field starting to play around with this in different ways, because as we all know, really having grounded responses or responses. That are relevant for your class.

It's not just factuality. It's also relevance taught the way that matters for your specific classroom in your region and your school. That's so important in education. People really want to be able to engage with these models in that way. So that opens up all kinds of different possibilities where you can have schools now, for example, being able to connect it with Google drive and all the resources that teachers may have and leveraging all of that grounding in, in, in these responses.

So it is still very early days on this, but I think it is one of those breakthrough capabilities that opens up a whole new set of possibilities. 

Ben Kornell: Okay, so my last question is really around Google is really the largest ed tech company in the world. I know it's probably a rounding error in the Google BNL, but really, there's a way in which more educators and more learners are connected on the Google platform, whether that be cloud, Chrome using classroom, using the work suite.

But that's also a great responsibility and many of these other, many of the startups or many, of the other AI large competitors are all thinking about, okay, how do we reach educators and younger users and so on? How does that affect your thinking as you lead Google for education?

How do you think about which tools to move forward with and infuse into the suite and which ones to hold back on or stay cautious about or just continue to have in the ice box that you're not going to push forward? How do you weigh that? 

Shantanu Sinha: Yeah, so we take that responsibility really seriously.

And I think it is something that we spent a lot of time with educators. We spent a lot of time with our customers really understanding what their needs are and ensuring that anything we roll out, we can do so in as responsible a way as possible. I spoke earlier a little bit about really understanding the technology and the use case and making sure you're getting it right.

And given our scale, we understand the obligation that puts on us. Just as an example on that, last year, there was a lot of talk about people plagiarizing with degenerative AI models, having them write your essays for you. And we have a feature as a part of Classroom that's called Originality Reports, which is basically a plagiarism detector that checks whether your work is original or if it's somewhere else on the web or somewhere else.

Another student in your school or district as has submitted very similar work. So we looked at this and we said should we look into this AI detection and I looked at the research and I looked at, what's possible here and very quickly we looked at it and said, no, that's like it would be very irresponsible to bring out an AI detector to our user base because the science shows you that these detectors are getting 5, 10, 15 percent false positives and these models are moving so quickly across all the different providers.

There's just no way you can put something out there that has that level of accuracy. That's so important in education. And for us, we will really look at the implications of that, right? If a teacher is falsely accusing a child of cheating, Yeah. One in 10 times or one in 20 times, that is absolutely not acceptable in education, right?

Like we, I do not do that as a responsible way to bring that technology out there. And it is a high bar. And we really do expect in many of these metrics when we are running evals that we are getting it. It's not just directional. They're like, Oh, we could do that, but it's really getting it to the level of quality that is expected because it's so important that we get it right when we bring an education.

So we built that trust with our products for many years. We take that really seriously. It's very easy to lose that trust. So it's critically important that when we do bring stuff out, we're bringing it right. We're bringing in a way that works for our customers, that works with their expectations, and really works with the unique needs of and expectations in education.

Because, as I was saying before, it is a very high stakes environment for a lot of people. So it's very important to get these things right. 

Ben Kornell: Yeah, with that, unfortunately, we have to wrap, but this is been over time with Shantanu for our EdTech Insiders subscribers. I hope everyone gets to enjoy this bit.

And, for those who want to follow up and learn more, Shantanu, you mentioned that you have a blog, what's the name of the blog and the address again? 

Shantanu Sinha: Yeah, the keyword blog is the Google blog, and we'll be posting all of our education announcements as a part of that. 

Ben Kornell: Awesome. Shantanu Sinha, GM of Google for Education.

Such a pleasure to have you on EdTech Insiders. Talk to you soon. Thank you.