← Building One with Tomer Cohen
Building One with Tomer Cohen
Building Khan Academy with Sal Khan: Mastery Learning, AI, and Education at Scale
00:49:36
Loading summary
Transcript56 lines
- [00:01]
Tomer Cohen
LinkedIn news.
- [00:07]
Sal Khan
Innovation ultimately is solving problems.
- [00:09]
It came back to my own personal pain point.
- [00:12]
It just takes a while to build that trust. The problem that I was trying to solve, it was all I was thinking about. You have to be obsessed with the human condition.
- [00:20]
I'm Tomer Coyne, Chief product officer of LinkedIn, and this is building one.
- [00:26]
Look, don't get a scriptwriter and a bunch of standards folks. Just get someone who's excited about the topic to explain it as they're talking to their cousins that they care about. Because that is literally what I was doing.
- [00:36]
That's Sal Khan, the founder of Khan Academy. He's talking to me about bringing the concept of minimal lovable product to how he builds great products. We're going to get into that and so much more right after this.
- [00:54]
Tomer Cohen
The LinkedIn podcast network is sponsored by Dell AI Factory with Nvidia, which provides AI solutions that are easy to implement, secure, and tailor to your business. Visit Dell.com to learn more.
- [01:11]
Sal Khan
If there was a home screen to represent the apps my family uses the most, Khan Academy would be one of the top ones right after Spotify. For the past decade, I've been using Khan Academy to tutor my kids. And I'm not the only one. Today, Khan Academy could be considered the largest school on Earth. That's why I'm excited to have Sal Khan here with me today. Sal started tutoring his cousin in 2004 to help her with homework. But 20 years later, Khan Academy is used by over 150 million learners worldwide. In this episode, we'll discuss Sal's principles for building powerful educational products. How he applies the concept of minimal lovable product, the craft behind building trust and motivation, the promise of AI in driving personalized learning. Plus why it's not a substitute for a human tutoring experience, and how aspiring founders should consider whether a nonprofit is the right choice for them. Let's get into it. So many people know your remarkable story, right? How you went from tutoring your cousin to being a family tutor to Suddenly tutoring over 150 million people around the world. Probably the most remarkable tutoring story of all times. Education has been a theme throughout your life, from what I've seen. But I'm curious how your career turned out differently than what you expected. Was education always there for you?
- [02:36]
I was always a kid that was confused about what I wanted to be, and it was usually a good problem that, you know, there was a lot of different things that I was interested in entering high school. I thought I wanted to be a cartoonist. I actually don't tell a lot of people that because it's so different. I guess I still draw things every now and then. Exiting high school, I thought I wanted to be a physicist, go to college, and I'm. I discover that even better than understanding reality is maybe constructing your own reality. And so I got into computer science. But even in undergrad there was about a semester or two that I thought I wanted to go into law, I thought I wanted to go into medicine. There was a year that I wanted to do a PhD in computer science, and I was still pretty confused all the way through my senior year in college. This was the late 90s. And then I saw how much tech companies in Silicon Valley were paying poor college students like me with a lot of debt. And so I started in tech, worked at a few startups, and then when the dot com bubble burst, I said maybe I should go to business school and try to figure out my life. And that's when I went into finance. So my life has been taking these twists and turns. I've always been super interested in education. You know, I used to say if There was a PhD in everything and if it really emphasize teaching, that's the thing that I would want to do. But there really wasn't that. A lot of the software projects I worked on, especially in college, were all around helping people learn. So I kept gravitating to this. And even in my hedge fund job, my now wife used to give me grief about people with a lot of education and capability. Just go and work at a hedge fund. Why don't you do something useful for society? And I said, oh no, no, I'm going to do this long enough and one day I'm going to start my own school because I have ideas, but I had no idea of how it would actually take me here.
- [04:16]
It's pretty amazing to see how the ducks connect, but it's really a convergence story more than anything else, you know.
- [04:21]
Yeah, in hindsight I can draw a narrative. If you met me in the late 90s, you said, that guy's confused.
- [04:28]
And I heard you talk about how you've always been fascinated with the role software can play in elevating human potential. Outside of education, were there other areas you were thinking about? Software and intersection of medicine and so on?
- [04:40]
In the late 90s, you know, I tried to do a couple of startups that never went anywhere. I worked for a couple of startups, I thought they were cool, but if I'm honest, my heart wasn't in it to the same Degree. You know, I was doing it for, hey, it would be cool to build some software that a lot of people use and maybe I can make some money. That was my motivation. But I've always been fascinated by human potential. Everyone has so much potential. Most people are well below it. And it did feel that technology was clearly giving access to knowledge. But how could we help people level up their own potential? Not just their potential, but then their access to opportunities just felt like the biggest problem to solve.
- [05:18]
You know, growth mindset is kind of our second religion at home. There's a great saying by Carol Dweck that it's not about meeting your potential, it's building your potential. Your potential ends when you die, in a way, but up until there, you're just building your potential up and really talks volumes to what you're talking about. I'm curious. When we talk to builders on this show, you could kind of see how there is, like a specific trait, almost like a superpower, that they are indexing on that is really unique to them. When you look back at your career, do you see the same for you? Like, when I was going through your material and learning, I saw a lot about that willingness to take the leap of faith where maybe others would not do. Does that resonate or is there anything else that resonates for you?
- [06:01]
Yeah, I think for me, if I get excited about an idea or if I think there's a way to solve a problem, and this is one of the powers of software, this wouldn't have been possible. I don't know, 60, 70 years ago, I was empowered enough to let me try this out, let me put some stuff together, let me prototype it, make a couple of videos. People forget this was before cell phones had cameras. And I said, okay, if there's some way I can capture this screen, et cetera, et cetera. And then, you know, I was learning about YouTube, which at the time was a new startup. But I have a bias towards action and just try to put some stuff together, see if I can do it. Or our team could do it later on and then put it out there and see if it resonates.
- [06:41]
You know, on that same topic, when it comes to building products, many indexed on the concept of mvp, minimum viable product. And it looks like this came naturally for you. Khan Academy is, in a way, like a masterclass in mvp. You started with phone calls and Yahoo. Messenger and Doodle. You shifted to YouTube and then you coded your own first product. And then you gradually built like a platform. This was kind of like brick After Brick, that notion of being scrappy, early to start nimble. Is that something you hold as a principle right now within Khan Academy?
- [07:16]
I try, and I wish I could say we always did the true mvp. And sometimes we say minimal lovable product is a word we sometimes use. You're right. Those early days of Khan Academy, when I was just scrapping together technology to tutor my cousins or make videos, those were very mvp. Although I think there's an interesting dimension where let's even take these videos, for example. If someone had a larger budget or they worked for a publishing house at the time they said make some educational videos, they definitely would have been more than what I was creating at the time. They probably would have been highly produced, maybe with graphics. They would probably have gotten a committee of people to write scripts and then a voice actor to read the script, et cetera, et cetera. But I think what was interesting about the way that I had approached it inadvertently was it was very MVP on the production value and on the technology. But I think it went beyond in many ways what some of the publishers would have done in terms of just emotive value. Look, don't get a scriptwriter and a bunch of standards folks. Just get someone who's excited about the topic to explain it as if they're talking to their cousins that they care about. Because that is literally what I was doing. Khan Academy, it went beyond the mvp. Not in the direction that most people assume you need to go, where it's more polished. It went in the other direction. You know, one of the things that I've always been a huge believer in, and I continue to be, is the power of personalization. Students learning at their own pace. But I realize even those early Khan Academy tools that I was creating, it was working for my cousins because I was there in their life, really motivating them to use it. But I think we were to some degree early in the market in terms of really giving students agency, really giving them personalization. Assuming, like, what if everyone had a tool where they can learn what they need and learn at their own base, Surely it'll fix all the problems. Well, the world is more complicated than that. But now, and especially as you're a larger organization, MVP or minimum lovable product, let's get something there fast, learn from it. Let's not spend a year to get something huge out there and then discover that that's not what folks want. So we try, but I think we also fall into the pitfall of over engineering pretty often.
- [09:21]
You made a profound point there. In reshaping the concept of mvp, I think for many people it was really the functionality you offer. But when I look back and again, I have hours of watching you tutor with my kids, there's cases when you're like, oh, let me fix that. You erase and you go back and in many ways you can say, let's actually edit this out. Why would we see the erasing? But no, that's actually part of the human experience. It's part of me building trust that there is a real person on the other side and we can all as learners go and make mistakes and just literally erase them. And in many ways, what's missing in MVP concept is the emotional attachment. It's not about the functionality. Sometimes I think this of the idea of like lovable product because it inherently brings emotion in, not just functionality in. There's something really profound about. It's not the production value, it's the connection you have with the learner, it's the emotion you're evoking, especially with learning, which needs to be more of a motivational thing coming from the learner itself.
- [10:22]
You know, I agree on that. It was an inadvertent thing that people would have considered a bug in those early videos that became a feature. And I sometimes wonder whether even we've lost some of that charm because now we do have to be more tightly aligned to standards. I do. If I make a mistake, I sometimes just rerecord the video. Hopefully we haven't over polished. But even when we talk about this, I'm like, I hope we don't lose that ourselves.
- [10:45]
One example of that that might resonate is I think after we had a lot of courses with you, and again, I do this with all my three kids and now they're 12, 9 and 6. I think we shifted to somebody else for a few lessons and it was so noticeable. She was great, but it was, hey, I built this relationship with my tutor. I know how he sounds, I know exactly how he operates. And it's not about having somebody else's grades. There's almost like a connection there. Is that something you're thinking about?
- [11:13]
Yeah. And I always try to keep my own ego in check on this because, you know the math videos on Khan Academy, I still make all of them and I make a large proportion of the science videos, even some of the humanities videos. But we do bring other folks in. But I've tried to stress to the team because it's not always obvious, especially in, I don't know, organizations. People just think, how do you Scale, how do you do things professionally? How do you do things in kind of tighter, more structured ways? And it's not natural for the CEO to continue to make the videos. People like, this isn't a scalable process. These other standards that we're missing, we could just get someone else to do it. And selfishly, I enjoy making them. I believe some of what you just said, but I tell our team, imagine if you're teaching a course and no matter how good the substitute teacher comes in, it's a substitute teacher and it just takes a while to build that trust and that rapport and how are they thinking, et cetera, you might not just be open to them, no matter how good they are. And so I do think there is a continuity of narrative that is very powerful. And I always feel a little bit shy, even in our own organization about pushing my own videos. We just had a conversation last week where they're like, sal, we got to make all these new videos for these new state and country standards. We don't think you have bandwidth for them. It's going to be 300 videos this year. Maybe we should look to hire. I'm like, no, no, I got them. I got them because I don't want to. One is oftentimes also you bring in external folks and they're going to want to write scripts, they're going to want to do this and that. And you know, the video that I can crank out in 20 minutes, they're going to take two days to make and then I'm going to spend 20 minutes editing it or giving feedback on it. And then it's going to feel like that polished thing from a publishing house that we didn't want to be to begin with.
- [12:53]
Yeah. In many ways I think it's almost like a principle of learning can be seen as a left brain experience, but it's really a right brain experience. You can learn when you have trust established. On the topic of principles, one of the ones I learned about Khan Academy is the idea of mastery learning is that you move on only once you comprehend the topic really, really well. Maybe it sounds for folks like, obvious, but it's not. Most traditional schools do not do mastery learning. They do. You know, you have enough to. It's sufficient to move on, but it's not about mastery. And this is quite unique to Khan Academy. I haven't seen this in other traditional schools or E learning platforms. Can you share why this was so important for you as a principal?
- [13:38]
Yeah, Well, I actually think, you know, the issue is that with Traditional schools, they don't even get to sufficiency. They don't have any threshold. They just say, you saw it, we covered it for a couple of weeks and now we gave you a test. And yeah, maybe for some students, you know, if you define, I don't know, let's call mastery 90 to 100%. It depends. Maybe sufficient is 80% that, like, yeah, you could start to learn other topics that build on it, but you haven't fully mastered this. But what we see is a lot of students might get a 70% on a test, or maybe they get a 90% summer happens and they forgot all of the material and then they build on that. And this isn't just theoretical. The majority of students in the United States, and this is true in most of the world, the majority who graduate from high school and then decide to go to college. So this is really about the top half of students, generally speaking, the majority of them, when they go to college, they take a placement test. And those colleges say, you're not even ready to learn algebra yet. You need to go back to essentially pre algebra, seventh grade math, which shows you that students were being pushed ahead year after year after year, taking classes called algebra 1, algebra 2, geometry, pre calculus, Calculus in some cases. And then the colleges say, wait, you're not even ready to learn algebra yet. So, yes, it's somewhat common sense that if you haven't learned your negative numbers well, you should learn it at least to a sufficient level. You know, mastery we can debate. My view on it is you have to learn to at a level that when you build on top of it, it's not going to be confusing or debilitating. And you should always have the opportunity, the incentive to revisit a subject in case you need to get to that level of mastery, fill in any gaps. You forgot some of it. It's kind of common sense in hindsight, but we've never done that historically in the school system because we didn't have the resources or the logistics or the technology to do it. Because if you're one teacher, 30 kids in a classroom, they're all at different levels of mastery. You're told by the state that you have to cover these 60 standards in 180 school days. You move on and you just hope that some of the students are going to be able to keep up with it. But the average American school, and I think once again, this is true internationally as well, has three to five grade levels of ability in each classroom. But we just pretend that there's one Ability, and we just keep pushing people forward.
- [15:43]
For myself, I didn't call it mastery learning, but for me, foundations was so important. I was very slow to learn basics. I took my time with the basics, but then once I learned the basics, it just felt like I was flying on the rest of the material. So I'm curious, what was it for you early days that kind of built this conviction around mastery?
- [15:59]
Yeah. You know, in the early days, I didn't have the language of, you know, the word mastery learning or personalization. I knew a couple of things. When you do well in school academically, there's one of two narratives you can tell yourself. You could tell yourself, oh, I'm just gifted and those other folks aren't. Or you could tell yourself, no, there's just something about the way that I approached the subject that just made it. That made it easier to absorb. And for me, and it sounds like you might have been very similar. And actually, I think most people who've been successful academically, especially academically in stem, when we were learning how do you multiply decimals or do long division? I'm guessing. I mean, I was. I could speak for myself, the kid that was like, well, why does this work? And I would still play with it. And I was like, oh, oh. And I could do it this way, too. And I could do it this way, too. And, you know, later, when you're learning algebra, well, why does this work? Why do I have to do things? Oh, because you're literally saying that the left is equal to the right. So if you have two things that are equal to each other, if you want them to still be equal, you have to do the same thing to both. And so I think that process of really building the intuition, and then later on, you're in physics and you see your friends trying to memorize that formula, it's like, no, that's just a rearrangement of the formula we learned last week, or that's just a rearrangement of that formula from algebra. And then when I saw my cousins when I was tutoring them back in 2004, 2005, 2006, I saw that's what was holding them up. They're hardworking, smart kids. Frankly, they're going to much fancier schools than I went to in New Orleans. But they were memorizing formulas, they were cramming for the next test. And I was like, you know, Nadia, which was the first cousin I was doing, I was like, nadia, let's just go back to the basics. This is very intuitive. What does that equation mean to you? Like, what is it telling you? And slowly but surely as she started to build that muscle of like, oh no, this is actually saying something and it should intuitively mean something. Then when you build on that, it just starts to make more and more common sense.
- [17:48]
Could not agree more. You know, I've been using Khan Academy with my kids for years now and I love the experience. But I also realized that it was far better when I was doing it with them because what I could do is I could personalize the material for them. It's like, you know, one of my kids, they love playing so I can talk about descent and one loves dancing and there's so many aspects. I could really bring the material to them in a much visceral because I know their hobbies, I know how they think about things. So when this new web of AI came out and with it came Kamingo, this felt like the beginning of a revolution. For the first time, personalized learning is doable. Is this what you're hoping for or is there a more ultimate kind of scenario you're hoping to achieve?
- [18:29]
Yes, and there's more and no. The one thing I've realized over the years, if I really introspect and what you just described I think reinforces that is motivation is everything. So when I was a kid and I wanted to learn something, I was a pretty motivated student. But the best, if I had access to a textbook, I would have to just power through that textbook and learn it from that. Maybe if I was really motivated I could go find someone to talk to. But that didn't happen too often. And the Internet and YouTube and Khan Academy has lowered the activation energy you need. It's much easier to learn calculus from Khan Academy than to learn it from a textbook, but it still takes a reasonable amount of motivation to engage. And so what we are seeing is yes, there's probably 10 or 15% of the population that's already motivated and already has a reasonable foundation and has the self confidence and they can go on Khan Academy and they're off to the races. And we have some incredible stories of young girls in Afghanistan. One of them, who's now a quantum computing researcher at Tufts, even though she was in a Taliban controlled territory and she wasn't allowed to go to school and she just self educated on Khan Academy there's an orphan in Mongolia who also did the same thing. So there are stories of these very hyper motivated people, but we realized for the majority of students, yeah, you need someone in your life or something in your life. I think the AI will help. As you mentioned, the AI can answer questions if you're confused, it can connect it. You can say, why should I have to learn this? You know, we are building ways that the AI can act more as your concierge, as your coach, hold you accountable. The way that I held my cousins accountable, the way I hold my children accountable today, that you're holding your children accountable. And this is something I don't say just because it's fashionable to say, I truly believe it. The human element is not going to go away. And if anything, as the technology gets better and better, hopefully it liberates the human element to focus on the human element. When a teacher is just lecturing in front of a room with 30 kids, that's not a human connection for anyone. But if the teacher can sit down next to students and say, hey, I really believe in you. You just got to do this. You just got to put in a little bit more time this week. Trust me, if you get to this and we just keep chipping away at it, in two months, three months, you're going to see a big difference in your math confidence or in your writing confidence or whatever. That's what makes a difference. And I see that with my kids. My kids, you know, they're different. My. My oldest put them on Khan Academy, and he just off to the races, and I'm like, oh, this is easy. My middle one, that's a little bit more coaxing, you know, working on her confidence, et cetera. My youngest one, you know, I'm not ashamed to say there's a lot of bribery involved. Just, you know, you get to 5%, you're going to get a Pokemon pack, you know, whatever. And actually, what's interesting, I had to do a lot of bribery last year, but that got him going and created the habit, and now he doesn't need the bribery anymore. He's seeing the benefits of just doing it on his own. But once again, you being in your children's life, me being in my children's life is super crucial. Hopefully, though, the existence of Khan Academy means that you don't have to be an expert in the subject matter. You just have to be an expert motivator. I do think that makes all the difference. And by the way, it's a great way to connect with your kids over something substantive. My oldest is taking an advanced physics class that we don't even offer it on Khan Academy. But I'm geeking out with him now on his problem sets. We're Talking about it. We're theorizing about, oh, what does this mean? Why is this? And it's great. Just bonding. And my first book, One World Schoolhouse, there's been all these studies during the Cold War about, oh, the Soviets are ahead, we need to have more homework, especially in stem or less homework. And, and there's no correlation between the amount of homework and success, but there is a correlation between having dinner with your family and success. And so I think the more that parents, siblings, grandparents, uncle, aunts, cousins, friends can really connect, especially over things that are academic. That's going to be a huge benefit for the students.
- [22:14]
I think also when you do Khan Academy with them, you do get to experience lows and highs. Also, it's a mirror to you about how do you deal when your kid has lows and highs. You know, you can unleash a high, but you know it's a fake high if you just give them the answer. And it's not pleasant to be in a low with your kid, but it's really important for learning. Like, there's, I believe there's no learning without friction. Like, you have to have some friction to learn. Like, your brain needs to evolve. It's not always fun to me. Actually, most of the time it's not so fun to me. But in retrospect, it's like working out. It's one of those where, like, it's always valuable. We're going to take a quick break, but don't go anywhere. When we come back, we'll learn how Khan Academy is working with school districts to improve outcomes.
- [22:57]
There were districts where maybe 30% of their teachers were by choice using Khan Academy. But what about the other 70%?
- [23:08]
Tomer Cohen
What happens when businesses have Dell AI factory? With Nvidia, years of patient data gets distilled in seconds. Investors know the exact moment stock should be traded and quality checks speed up so production lines never slow down. No matter what business you're in, there is an AI solution that's easy to implement, secure and tailored to you, that can keep your business moving forward. Dell AI factory with Nvidia, your way to AI. Visit Dell.com to learn more from the minds of visionaries.
- [23:42]
Sal Khan
The desks of disruptors. Hi, I'm Lars Schmidt, host of the Redefining Work podcast. Join me each week as we explore the new world of work through the lens of those shaping it. CEOs, HR leaders, investors, and more. Be a part of the conversation that changes everything. Subscribe to Redefining Work today.
- [24:09]
Okay, we're back. And I'm speaking with Sal Khan, founder of Khan Academy. Curious. When you talk about Kamingo or maybe just in broader AI, this is in a way uncharted territory. So, you know, obviously one trait that comes from you loud and clear is the motivator. But when you think about, I don't know if I would say the personality of what you're trying to build, is it the same personality of Brand Academy, which is supportive? I think you guys have quirky in the brand as well, which is awesome. But are you thinking about the AI agent differently? Like how do you think about the principles for how you build that personality?
- [24:44]
We're all figuring this out as we go, but my bias and our organizational bias is to give Khanmigo, at least at a default state, a lot of the traits that hopefully made Khan Academy videos and even early exercises useful for folks. And yes, a little bit of quirkiness. Plain language. It's amazing even today that the default ChatGPT or Gemini is so dry and just, you know, if they just wrote like a two sentence basic prompt behind the scenes for it, it would be so much more engaging. But they just don't take the trouble to do it. And we've taken a lot of trouble to try to be in the right zone of, you know, it's not talking down or diluting the content, but it's not giving you these big Wikipedia type explanations. It's being Socratic, but still helpful. It can be fun, a little bit quirky, uses emojis, et cetera, et cetera. So I think there is a lot of work other than just making the quality of the information as good as possible, but making how it's delivered as engaging and as fun and as quirky as possible. You know, we don't exactly know what AI is going to be capable of in five years, but I can imagine that you're going to be able to have a video conference with an AI like the way we're chatting right now. And I would love that AI to yes, be very supportive. Quirky, focus on intuition. Now we will be able to give people the option of multiple personalities. A B test. Maybe there's one personality that works better than another. Just as an experiment, I tried a Bruce Lee personality once and I loved it. I loved the Bruce Lee personality for myself. When I ask a question, they're like, you know, why are you relying me on the answer? When you have the answer in yourself and it would say things like that, I'm like, you're right.
- [26:17]
Ancient wisdom.
- [26:18]
Ancient wisdom. So, yeah, I think we'll try out different things and see if there's some that universally work better with some folks or maybe people can choose.
- [26:26]
You mentioned the idea of having it as a principle being Socratic. Was this just watching you teach and, like, basically having it learn from your videos? Is this just like going for books and books about, you know, how to teach kids? Somebody who's building that, it's not easy to build it. So, like, how do you go about building that?
- [26:44]
It's something that I try to force myself to do, especially if I'm, you know, we have a lab school out here in Mountain View Con lab school. It's where my kids go. And I work with especially the high school students pretty frequently. And, you know, I always remind myself before I enter the room with them, I say, sal, ask questions, don't give them answers. You know, every now and then, if they're really stuck, maybe I'll say, hey, have y'all thought about this or that? But for the most part, ask questions. And it goes to your point that learning happens through friction. Learning happens through using multiple parts. Not just listening to something, but try to construct your own responses. Try to draw the connections yourself, Try to make sure you can listen to other people, which is just a very important life skill, and keep a continuity of the conversation as opposed to just jumping from here to there. And so, yeah, that was the spirit of Khanmigo. And, you know, we are having debates inside the organization. Like, we have a default Socratic tone where it's going to ask you questions because we don't want it to help you cheat. But we do know there's cases where students want just a simple explanation. And right now, they might just go to chat GPT because they don't want to necessarily have this, like, you know, Socratic. But then we're like, well, what about cheating and all that? Well, what if we let parents and teachers monitor it? Or what if we can have the AI monitor what's likely cheating behavior, not cheating behavior. We do have these tensions, but I do think all else equals. Socratic interactions are definitely the best way to learn and to retain the information.
- [28:04]
Zooming out, there's kind of a holy trinity of, like, the child, the teacher, and the parent. And if they're all in sync, it feels like it's an incredible experience for the kid. And if they're not, you could see how the learning experience just goes sideways. If one of those nodes do not play along or they're not in sync, are you thinking of each one of them when you think of Khan Academy, they're all contributing in a way, right?
- [28:30]
You're right. Those three can work in harmony. Outcomes are going to be good on multiple levels, not just academic outcomes. I think emotional and other outcomes will be positive. So we have already tools for obviously the students and the teachers and also for parents. But having scarce resources, we always have to pick our battles. And I think the battle that we really want to nail, and we haven't nailed it yet over the next several years, is that teacher, student. And actually there's another stakeholder, the district or the system that you also have to bring in alignment as well. This goes back to if you have a family that's pretty educated and they have resources and they have time and interest, then they'll do what you're doing with your kids, what I do with my kids. And kind of regardless of what's going on in school, we can provide a safety net of where they're going to have solid academic foundation. But that's unfortunately not necessarily the majority case. Every parent wants the best for their child, but they might not know how to navigate the system. They might not have the time, they might not have the knowledge to tutor their kids in algebra or world history or whatever it might be. And so that's where the teacher is super important. And we've seen that we have a huge emphasis on tools for teachers, AI tools, including help them with lesson plans, et cetera, et cetera. We have a whole project where the whole loop is very integrated. You know, the district can give the AI context of curricula and other things, and teachers can use that and customize it, their lessons, et cetera. They can create quizzes, they can use existing Khan Academy practice material, administer it to students across subjects, not just math. I mean, we have some writing coach, we have reading comprehension, we have history, science. And then the AI can build inferences and then use that to be as a teacher's aid for the teacher to customize the lessons and keep rinsing and repeating, letting the teachers know what students they might want to spend a little bit more time with because they haven't gotten to at least a sufficient level of proficiency. And so they're just going to fall farther and further behind. But if we can really start to streamline that, I think that's going to help a lot. A lot of kids that right now aren't getting that type of support.
- [30:33]
You mentioned the district side. It always felt like one of those areas that if you solve, can help you scale, but also can slow you down. From scaling because they're bureaucratic, they're slow. Especially with innovative products like Khan Academy. It's not necessarily like, let's go after it. It takes time to build with the districts. I'm curious, when you think of those constituents, from teachers to districts, what challenges they're dealing with that you're like, okay, if I solve this for them, I'm really hooking myself into the system in a way that helps them accelerate versus becoming one more thing they have to deal with.
- [31:06]
Yeah. About seven years ago, almost eight years ago, Khan Academy already had a ton of efficacy studies. We already had a lot of users, A lot of teachers were using us in their classrooms. And then we say, well, you know, how come districts aren't using us more systemically? Like, there were districts where maybe 30% of their teachers were by choice using Khan Academy, but what about the other 70%? And so we started reaching out to those chief academic officers, superintendents, and pretty much every conversation was the same conversation. We would say, hey, you probably heard about Khan Academy. Here's our efficacy studies. But, you know, I heard y'all are either not using some tool or you're using that tool and you're paying a lot of money. And they have no efficacy studies. And no one likes that tool. It's coming from a publisher. It came with your textbook, whatever. And almost every school district leader said, no, I agree with you. In fact, when my niece knew that I was going to speak to you, she said, thanks for getting me through the sat or, you know, now there's even some young superintendents. You got me through grad school, you know, stats on Khan Academy. So I believe you, I believe your efficacy studies. But if you really want us to systemically use you, you need to give us support, training, integration with our rostering systems. You have to check some boxes on some of these regulations that we have to meet. That was when we said, look, if we're really serious about our mission as a nonprofit, free world class education for anyone anywhere, if we're really serious about reaching all students, you know, this didn't feel as much fun as just making cooler products and cooler content. But this is something that we had to build this muscle. And so we did start building that muscle. And we've been building it ever since. And some of our fears have manifested that it is sometimes bureaucratic and you sometimes have to jump through hoops that feel not like a good use of resources. But, you know, we have about a million students in these formal district partnerships now, and they are really using Khan Academy at a much deeper level. And there's a few places, Newark, New Jersey is one of them. It's a large urban school district. They're seeing real gains, not just efficacy study gains, like district test scores, moving ahead and just better engagement. I met a young boy in Newark who's in third grade and doing algebra two now and would have never even had that opportunity. People even didn't even know he was capable of that. And so, yes, I think it was a good decision. And especially now in this AI world, I think being able to fit in with the system is actually where we can add a lot of value in our team. And I think every company is like, well, ChatGPT and Gemini and Anthropic. They're just going to get better and better and better. And at some point, if someone is suitably motivated, they'll probably just be able to Give the video ChatGPT a phone call or chat with it. It'll tutor them and whatever else. You could already prompt it to be a okay tutor. But the real value is how do you use the AI to streamline and provide information in the cycle between the districts, the teachers, the students, to your point, the parents, maybe there's ways of providing signals of what students know to say, colleges, et cetera. And then how do you do that in a way that is pushing the envelope, but it's safe, it's ethical, it's not biased, et cetera, et cetera. So that's where we've really leaned in.
- [34:07]
When you talk about the experience with these teachers, there's obviously like, if you fully integrate with us, you get all these amazing things and then there's. But if you don't, it's like you have all these tools and they're trying to complement their own curricular With Khan Academy, that sounds like very hard to deal with or like try to solve for. Is this something you're spending any time thinking about? Because that's kind of like the one minus if you integrate with us scenario.
- [34:30]
Oh, it's what we spend a lot of our time thinking about. And you're absolutely right. The reality of your average teacher right now is the district has adopted some curriculum and put Khan Academy aside. We have one of our funders said, why are you creating these AI lesson planning tools? Haven't districts adopted curricula? And I was like, well, go talk to any teacher. Yes, they've adopted something, but the teachers don't want to use it exactly as it's written. And the Average teacher spends 10 to 15 hours a week actually planning lessons. So kind of either ignoring the mandated curriculum or just saying I need to modify that because it's not what my students need. So they're dealing with that. And part of it is they say, well, my curriculum doesn't really give a lot of practice, so maybe I'll use Khan Academy for that. Or my curriculum doesn't really help me make engaging activities during class time. So I might use Kahoot or bluekit or nearpod for something like that, or I use this tool for writing and that tool for math. So the average teacher has like 10 different tools with 10 different dashboards, all of which are supplementary to their core curriculum, even though they might lean on these supplements more than their core curriculum. So even the taxonomy I think of core and supplement is somewhat broken. I think AI is bringing in this opportunity of hopefully defragmenting it. And this is what we're trying to work on, which is we already have lesson planning tools. It's only a matter of time before AI can help. Even grading papers, things like that. People know us for math, but we're going in all of the core academic subjects. And so yes, a world where the districts can give whatever context they need, teachers can give the context, but the AI works with the teachers to customize, modify it, the way the teacher feels good about. And then all of these different types of modalities, whether it's in class games or independent practice, whether it's in math or reading or writing, etc. If it can be one stop shopping from a trusted provider and the AI is really providing insights and streamlining it, then I think that could be a win. But that's a bit of a bet that we're making.
- [36:29]
That'll be an incredible unlock. So famously, Khan Academy, you started as a nonprofit from day one, and I recall you talking about how you were clear about the value and the product you wanted to offer, but the business model wasn't so clear, which makes sense from a nonprofit perspective. What advice today would you give entrepreneurs or builders who want to build similar mission first paths and how sustainable is that in the long term and what they should think about right now.
- [36:56]
In retrospect, we'll see how sustainable Khan Academy is in the long term too. But for the most part, I'm very happy. The early days of Khan Academy, I was always worried about it being a flash in the pan. And now we're 16, 17 years in. Our scope has grown, our budgets have grown. I get anxiety every time I think about how much money we have to raise but there's also our legitimacy has grown, and so that helps us raise those resources. Even though in some ways Khan Academy and myself were a bit of the poster child for nonprofit tech intersection, I would not just immediately go in that direction because it's not easy. It's going to be harder to raise money, it's going to be harder to get some of that initial legitimacy, et cetera, et cetera. I would ask, could you do this project in the market and will the markets function properly? If they do, and if you can do it well, then I think a for profit could be a completely reasonable way to go. You know, before Khan Academy, I worked at a hedge fund as about as for profit as possible. But my thinking is, when do you need nonprofits? You need it when the markets are either not functioning properly or they're functioning in a way where the outcomes are not consistent with our values as a society. You know, I think that this happens probably most often in education and healthcare, where, like, if someone's sick, they should be healed, we think. I think if someone wants to learn, they should be allowed to learn. We all benefit. If they can tap into that. You could say, well, is that the role for government to just make sure that these things happen that way? Well, we know government maybe in the very long term, but they tend to be slow and bureaucratic and political, and it's not going to happen. I remember in the very early days of Khan Academy, this was like 2000 before I quit my job. 2007, someone from the state of California had reached out to me and said, oh, we're gonna create a California Academy kind of thing. I mean, they were essentially describing Khan Academy and they had a huge. But I forgot where, you know, it was like 20 million or 50 million. It was some huge amount of money.
- [38:51]
Just give it to me, I'll figure it out.
- [38:52]
Well, that's what I was thinking. I was like, they're like, oh, how do you make your videos? I'm like, well, you know, I could do this for you. I'll do it for you for like $200,000, not 20 million. And so the idea was there, the intent was there, which is even better than actually many governments would do. But it never went anywhere because it was bureaucratic, it was political. They don't know how to execute. And now with the pace of change, even faster. I wouldn't depend on government.
- [39:16]
Could you ever see taxes covering this type of work? To your point around, like, everybody should be able to learn and get access to health care and yeah, look, one.
- [39:25]
Model is, I kind of see as SpaceX is to NASA, Khan Academy can be to the education system. You're still going to need maybe some form of government agency. And look, now that we're working with school districts and these school districts, when they ask for support training, they're asking for someone to go out there and do professional development. They're asking for some bespoke integration with their IT systems. We have to charge for that. We charge a very small amount, $10 per student per year. So, you know, these districts are spending $30,000 per student per year and we think this can accelerate them pretty dramatically. But that is, at the end of the day, taxpayer money, some small fraction of which is coming back to us as a not for profit.
- [40:01]
I love the idea of Khan Academy as a rocket ship for education.
- [40:05]
Well, it's right. I mean, you know, my family in Houston, we were visiting the Johnson Space center and you know, it's great. They were showing some of these other rocket projects like this Orion launch system from Boeing, et cetera. I think they said they spent $25 billion on it. So far that still doesn't exist, really. And it's going to cost $2.4 billion per launch. And then I just randomly took my phone, I'm like, I wonder what, what the starship launches cost, which already exists. It's going to have higher capacity, blah, blah, blah, blah, blah. And already it's sub $100 million and Elon wants to get it below $10 million a launch. So it's literally like it's 1 20th today of the theoretical cost that Boeing is saying. And it's going to be 1, 1 200th and maybe 1000th at some point. And I don't even view that as competition at some point. And I view, you know, if you said SpaceX, Boeing, I would say Khan Academy, traditional publishers. But hopefully we can play a role like that where we can nimbly and very efficiently make these systems that otherwise wouldn't be made based on traditional market forces.
- [41:06]
Sal, this was wonderful. I feel like we learned so much about how you think and how you build and some of the critical decisions you made along the way about building what it is today, I think the largest school on Earth. Can I say that?
- [41:18]
Sure, depending on how you define school.
- [41:21]
Well, school in my books. Thank you so, so much. There's a lot of great takeaways, so let's just dive right in. First, many of us think about minimal viable product and Khan Academy Start is probably one of the best examples of that. But over time, Saad has learned to think differently about it. It's less about viable and it's more about lovable, minimal lovable product. The best products establish a relationship with the user. I loved Sal's example about how in his videos he would sometimes need to erase something and correct himself, but instead of editing it out, he actually kept those moments in that authenticity can build trust. It's not about how polished your product is, it's about how it evokes emotion and builds connection Next, the principle of mastery learning is core. To Khan Academy, the focus is on building a solid foundation before expanding your knowledge. This is a key difference in how Khan Academy teaches versus the pass fail standard in traditional schools. It started with math and focused on creating a robust learning platform for it and only then scaled to other subjects and tools like Kaminga. The lesson here is don't rush to scale before ensuring your core is strong. This is true for your knowledge and for your product. Third, scaling a product isn't just about adding great new capabilities for your users. In edtech, it often involves integrating into larger complex systems like school districts and different state guidelines. That requirement shifted Khan Academy's efforts from not only innovating with their consumer product, but also developing the operational capabilities required to scale. While this might seem less glamorous than product development, it enabled Khan Academy to scale their impact and support millions of additional students. Fourth, for sal, AI is about personalization at scale, enabling a tailored experience for each student not only in content, but also as a coach that holds students accountable and guides them through the learning journey. SAL also sees AI as a way to free teachers to focus on high value human centered activities like motivating students and building trust with them. AI can also streamline the fragmented systems that many teachers navigate today. It can consolidate lesson planning, assessments and content in one cohesive platform. This is a dream come true for every student, teacher and also every parent. Lastly, Khan Academy operates as a nonprofit, but that is not necessarily a path SAL recommends to every purpose driven founder. In sal's view, non profits are most appropriate when markets do not function properly or when they function in a way where the outcomes are not consistent with our values. To him, everyone should be able to access good tutoring and reach their educational potential regardless of their socioeconomic status. When you think about this last takeaway, where do you see markets not functioning properly today? What nonprofit should exist to close those gaps? Feel free to let me know on LinkedIn.
- [44:34]
D
We'll be back in two weeks with Justin McLeod, the founder and CEO of Hinge. If you liked this episode, don't forget to rate and review us on Apple podcasts. It'll help people discover the show. Building One is a production of LinkedIn News. Our host is Tomer Cohen, LinkedIn's chief product officer. This episode was produced by Max Miller. Our associate producer is Rachel Karp. We're engineered and mixed by Asaf Ghidrang. Our editorial associate is Ally McPherson. We get additional production support from Alicia Mann at LinkedIn News. Sarah Storm is senior producer and Enrique Montalvo is our executive producer. Dave Pond is head of productions and creative operations. Maya Pope Chappelle is director of content and audience development. Courtney Koop is head of original programming. Dan Roth is the editor in chief of LinkedIn. If you know a product leader we can all learn from, send us a line@pitchesinkedin.com.