← Return to Index Archived September 14, 2026
The Lead — Sep 14
WORKLIFE WITH MOLLY GRAHAM · TED

Can AI really teach your kids? with Leah Belsky

OpenAI’s Leah Belsky argues that AI may broaden access to learning and give students new power to build, even as it cannot replace the human work of teaching. The real challenge, she says, is helping schools distinguish between information on demand and the productive struggle that makes knowledge stick.

49m / September 14, 2026 /aieducationtechnology / Transcript sourced from openai
All episodes from Worklife with Molly Graham →·Listen on Apple Podcasts →

Overview

Molly Graham speaks with Leah Belsky, OpenAI's head of education and a former Coursera leader, about whether AI will finally deliver on edtech's long-standing promise of personalized learning. Belsky argues that AI is already changing education because students adopted it faster than schools did, but she rejects the idea that it can replace teachers or eliminate the need for struggle in learning.

The conversation centers on a harder question than whether students should use AI: what kinds of learning should remain difficult, and where can AI help students create, solve problems, and gain agency?

Key Takeaways

  • Earlier education technologies expanded access but often failed to produce sustained learning. Coursera made elite university courses available worldwide, yet many learners did not finish courses and employers continued to treat degrees as a major hiring filter. Its larger impact, Belsky says, was giving people confidence that they could learn and improve their prospects.

  • AI differs from prior edtech waves because students adopted it from the ground up. Schools usually decide slowly whether to purchase and roll out technology. With AI, students were already using it before many institutions had policies or teaching plans in place.

  • Personalized answers are not the same as personalized tutoring. AI can explain material in different ways and respond to a student's interests, but a strong tutor also sets goals, spots misunderstandings, creates accountability, and makes students wrestle with ideas. Current models do not reliably provide those human elements.

  • Productive learning requires friction. Belsky compares AI to a calculator: using a tool before understanding the underlying skill can prevent learning, while using it after mastering the basics can support more advanced work. Copying an answer from ChatGPT may feel educational, but it can short-circuit the reflection and effort that make knowledge stick.

  • The more promising near-term use may be student agency rather than universal AI tutoring. Belsky describes students using AI to study possible treatments for bipolar disorder, make games accessible to blind students, redirect unused cafeteria food to food banks, and build small businesses. AI lowers the barrier to turning an idea into a prototype.

  • Teachers still matter because learning is social and emotional. They motivate students, understand their context, bring groups together, and decide when technology should be used. AI may become a better tutor, but Belsky does not expect it to replace the human presence of a teacher.

Practical Steps

  • Set explicit rules for AI use by task. For writing practice, math fundamentals, or other core skills, require students to work without AI first. Explain that the hard part is often where learning happens.

  • Use AI after an initial attempt. Ask students to show their draft, reasoning, or solution before they ask AI for feedback. Prompt the tool to ask questions, identify gaps, or offer another explanation instead of supplying a finished answer.

  • Turn AI use toward building. Pick a real problem at school, work, or home, then use AI to research it, outline a solution, make a prototype, or build a simple website. Repeat the cycle: identify a problem, make something, test it, revise it.

  • For educators, use AI to reduce administrative work such as drafting materials, organizing assessments, and handling grading support, while protecting more time for direct work with students.

  • Teach learners to distinguish information retrieval from learning. Ask: "Can I explain this in my own words? Can I apply it in a new situation? What did I struggle with before I understood it?"

Notable Quotes

  • Leah Belsky: "The struggle is where you're learning. So don't pull out the tech and don't pull out Chat."

  • Leah Belsky: "What we want humans to be able to do in a world where intelligence is infinitely accessible."

  • Leah Belsky: "They're just building. And that's what makes me so hopeful."

It may feel like learning, but taking away the friction, tears, and struggle can take away the very things that are so important to learning. — From the episode

Full Transcript

Source: openai 49m runtime

This episode is brought to you by LinkedIn. Running a small business means every hire matters. A bad hire can cost you time, money, and momentum. A good hire, they can change everything. But finding great talent isn't easy, especially when you don't have the time or resources to sift through piles of resumes and find the right fit. That's why LinkedIn built Hiring Pro, your new hiring partner that screens candidates for you, so instead of sorting through applications, you can spend your time talking to candidates and finding your next great hire. With Hiring Pro, you can hire with confidence, knowing you're getting the best talent for your business. In fact, LinkedIn found that its users are 24% less likely to need to reopen a role within 12 months compared to the leading competitor. Join the 2.7 million small businesses using LinkedIn to hire. Get started by posting your job for free at linkedin.com/worklife. Terms and conditions apply. This episode is sponsored by Range Rover Sport. Life moves fast, and when the world around you feels like it's changing a mile a minute, you need a vehicle that helps you rise to the challenge. Dynamic by design, the Range Rover Sport combines ultimate luxury and unbridled agility for a powerful drive filled with the latest innovations to keep you and your vehicle connected. An elegant 13-inch touchscreen lets you seamlessly navigate and control vehicle systems. And interior refinements like heated seating with a massage function mean comfort and luxury for every journey. Every detail of the Range Rover Sport has been engineered for impact, and its uncompromising design commands attention wherever it goes. With a variety of unique colors, interior finishes, accessories, and even wheel options, the ways to personalize your Range Rover Sport are nearly unlimited. When your time to lead arrives, you need a vehicle that rises to meet it: the Range Rover Sport. Exclusive offers available now. Explore further at rangerover.com. This message is brought to you by Apple Card. It's a new school year, which means yet another back-to-school shopping list. Luckily, when you use your Apple Card, you can earn 3% daily cash back on everything at Apple, including the new iPad Air and the MacBook Pro. Apply for Apple Card today and unlock the tools for a productive school year. Subject to credit approval. Apple Card is issued by Goldman Sachs Bank USA, Salt Lake City Branch. Terms and more at applecard.com. You know, what's really just struck me is I've been on this journey since starting in OpenAI to figure out, like, what do students actually do with this technology? Because the narrative out there is that they cheat. But what you actually see is that so many students, when they learn how to use it, are looking at sort of problems in the world that they want to solve and building solutions and not waiting for permission and not looking to hire engineers or VCs or networks or any of these things you and I had to do and probably still do. They're just building. And that's what makes me so hopeful. One of the side effects in working in tech for a long time is that you get a little cynical about revolution. Or at least I do. We've been promising that technology was going to revolutionize education for a long time. Online learning, MOOCs, smartphones. Then COVID forced everything remote. I feel like the main thing we've learned is that changing education is slower and harder than anyone wants it to be. So when people talk about AI transforming learning, my honest first reaction is, sure it is. That said, there is one thing that feels different to me this time. For the first time, it feels like we might actually be getting close to what edtech has been promising for 20 years: a genuinely patient, personalized tutor. Not an app that adapts a quiz, something that can answer your question, notice when you're confused, explain it to you three different ways and never get tired of you. And if that's true, it feels like it could change something fundamental, not just in schools, but in the way we learn. And it makes me wonder what we'll gain and what we'll lose. My guest today is Leah Belsky. She has spent almost 20 years at the intersection of education and technology. She helped lead Coursera during the rise of online learning, and she's now the head of education at OpenAI. She's been close to every wave of this changes everything in edtech, which means she knows exactly where the hype has fallen short. I want to know if she thinks this time is actually different, and if it is, what does it mean for us and for our kids? I'm Molly Graham, and this is WorkLife, where we untangle the messy human side of work. Well, first I get to start with my favorite part, which is, Leah Belsky, welcome to WorkLife. Thank you, Molly. It's great to be here. I know. I'm so happy to have you here. I've been really looking forward to this conversation because I get to learn all the things that you've been thinking about in the last year or two. Okay, so I just want to start, like kind of go back in time for you. You've been working at the intersection of education and technology for, I think, almost like 20 years. I want to know why you went into edtech in the first place, and what did you think was broken, and what were you hoping to fix? So I actually never intentionally went into ed tech or education, though I guess you could say that's where I am now in a certain sense. But it actually started back in college. I went to Brown and I was an activist, and I remember coming to campus. It was during the corporate reform movement, and had this idea with a set of friends to build an online investment club because it was a time where lots of people were trying to figure out how do you get involved in companies and create change. And there was this older student who was a successful entrepreneur, and he's like, Why are you going to build a club? Let's actually create a company together. And I remember during that time, you know, we then went to the CS lab and we searched for an engineer and we found some creatives. And eventually I have these vivid memories of driving down Route 95 in Boston, going to pitch some of the largest VCs, and it just made a huge impression on me in that I was like, Wow, I can't believe I, as a young person, can create and build something and sort of pitch some of the most powerful people in the world. And I think from there, I sort of got on this kick of, you know, this is what I want to do for my career. I want to enable as many people as possible to expand their capacity to be their best selves, to enable them to create and build, whether it be through coaching or technology or through education. And so that's kind of the, that's what motivates me to figure out how I can use technology and companies and organizations to help people be their best. But like, what did you learn from that early experience that you wish more people knew about sort of like what's possible when you use technology? It wasn't so much about technology, but what I realized is that if you can unlock people's confidence, unlock their connection, unlock their skills, and unlock their ability to learn. You can do wonders. So to sort of continue the story for a little bit, I went to the World Bank there after college that was very focused on international development. And then eventually was called by Coursera. Coursera, you know, some people know it, some people don't, but it's an online course platform that was one of the first to actually unlock access to some of the best university courses in the world. And it was there where I realized this sort of deep connection between this passion that I have of, you know, unlocking people's potential and education, because this was a platform that would allow anyone anywhere to actually start learning. And it happened to be a tech platform, right? So it really, you know, the passion wasn't for tech. The passion was for unlocking learning and potential. The idea was basically access, essentially. Like, what was Coursera really trying to do for people? You know, before platforms like Coursera existed, if you wanted to learn from the best professors in the world, you needed to get into college, you needed to pass an admissions bar, you needed to pay a lot of money for expensive tuition, and, you know, then you would get access to all those resources. And what Coursera did is it unleashed the university. So whether you were sort of a high school kid in India or whether you were a community college student here in the U.S., you could learn Python from the best professors in Michigan or any university in the world. And that was transformative. I think today, in a day where we have so much content on YouTube and there's a million online courses that might seem sort of part of normal life, but until platforms like Coursera existed, if you wanted to learn with a university remotely, there were some organizations that would be sort of sending textbooks and test materials across boundaries, but there wasn't much in access to high-quality education. Part of what I hear you talking about when you're talking about what Coursera was trying to do was kind of like break down walls. You know, that there were all these walls for people to access. Information and sort of like the best, right, the best professors, and that by breaking down those walls, you actually were creating more access to opportunity and unlocking people's potential, like you said. So when you worked there, what did you hope was going to change, like about the world and about, I guess, education? And what actually changed? Look, I think the big hope with Coursera was that, you know, you would have people from all the world who wouldn't necessarily need to pay for an expensive degree in order to get skills they needed to enter the workforce, and that it would essentially sort of open access to jobs and create opportunity. And, you know, and some of that happened, but there were a couple of big problems with the model. The first is that very few people actually completed courses. So a lot of people poked around and were excited, but didn't go very far. And then the second thing is that, you know, in much of our society today and in much of hiring, there's still a big filter out there for employers when they look at candidates, and, you know, the degree ends up being really important. But what I did take from the experience is that, you know, we spent a huge amount of time surveying our learners and really trying to understand what they were getting from Coursera. And the biggest thing that stood out is that Coursera, whether students did part of a course or all of a course, was unlocking people's confidence and giving them the sense, I can actually learn. I have a pathway to improve my life. I can learn IT skills and get a job at, you know, Google or wherever it might be. And That though they might not have been deeply learning skills, it was igniting their sense that they could learn something in the world, and that in itself was powerful. That's cool. Yeah, like the idea that just showing yourself that you can start something probably gives you then the confidence to go, if a degree is standing between you and what you want, maybe go get that degree. What did that era teach you about what's hard about change or, like, what doesn't—you know, you sort of have all these ambitions when you're in these companies, and, like, what did that era teach you about how education changes and how it doesn't? It was, one, going after the idea of allowing individuals outside of the education system to chart their own internal journey while also working with some of the best universities in the world and trying to change their business model. And so I think the biggest thing I was left with was how powerful a model that can be of change by allowing students to interact directly, but also trying to transform institutions. But also the other thing it left me with was just how challenging it is to drive change in universities and in the formal education system. Why do you think it's so hard? I think it's a number of factors. I mean, in the university system in particular, you know, part of it goes to the origin of these institutions and what they're designed to do. They're designed to produce knowledge. They're designed to produce research. They're designed to sort of put that knowledge out in the world. And along their way, they're also designed to teach students. And it is hard to accomplish all those goals at once. They are designed and structured with many, many different incentives. If I were to sort of start a startup and said I wanted my startup to produce research to disseminate knowledge and build world-class teaching, you as Molly would be like, No, no, no, no, no. What is your primary goal? Pick one. Let's get product-market fit and go. And so that's part of the challenge, sort of the many goals that these institutions are trying to solve and the various incentives they have in their business models. This episode is powered by AT&T Business. If you're running your own business, you know that the Sunday scaries are real. You're thinking about the invoices, the client calls, the inventory, the list that never actually ends. And honestly, we usually don't think about our internet or our phone service until it isn't working. Sometimes a simple glitch can feel like a total catastrophe. That feeling of powerlessness when your tech fails you is the worst. It's a reminder that our connectivity isn't just a bill we pay. It's the backbone of everything we do. AT&T Business gets that. They know that for a small business owner, good enough isn't good enough. You need reliability that stays in the background so you can stay in the lead. If you're tired of the friction points and just want a provider that understands the stakes, check out AT&T Business. Powered by AT&T Business. Built to work. Get AT&T Business at business.att.com. This episode is brought to you by Quince. Fall has a way of slowing us down in the mornings. Crisper air, hot cups of coffee, and a coziness you don't want to rush. But when it's time to get moving, I lean on pieces that make getting dressed easy so I can get out the door and still look great. That's what I like about Quince. They make quality basics designed to work across your wardrobe, everything from cashmere sweaters all the way to breathable activewear. I grabbed their micro rib tees and tanks in a few colors, and they've been the perfect thing to throw on with any outfit. Quince uses premium materials like organic cotton and European linen, and somehow everything is still priced 50 to 80 percent less than similar brands. Find the fall pieces you'll reach for most at Quince. Download the Quince app for app-exclusive offers or go to quince.com/worklife. Get free shipping on your order and 365-day returns. Now available in Canada and the UK too. That's Q-U-I-N-C-E dot com slash worklife. With no fees or minimums on checking accounts, it's no wonder the Capital One bank guy is so passionate about banking with Capital One. If he were here, he wouldn't just tell you about no fees or minimums. He'd also talk about how most Capital One Cafes are open seven days a week to assist with your banking needs. Yep, even on weekends. It's pretty much all he talks about. In a good way. What's in your wallet? Terms apply. See capitalone.com slash bank. Capital One, N.A., member FDIC. You've seen so many different phases, like online learning and Coursera's phase and mobile learning and remote learning during COVID and now AI. Like, so help me calibrate, like how... Help me calibrate, like, how different does this moment actually feel than the other waves that you've lived and led through? This moment is pretty darn different. You know, in the past waves of education technology, so I've worked in video when sort of video first came into the space. I worked at Coursera when sort of online learning blew open the doors of the university, and now we're focused on AI. And historically, when ed tech has come into educational institutions, it's been pretty top-down. So universities or schools get to sit there and think about, do we want to use this? Do we not want to use this? What's our policy? Should we train our teachers? And, like, slowly, methodically, over sometimes years, you deploy a technology. AI is grassroots. It's driven by students. They were some of the earliest adopters. At the same time, you have these students who are using the tech and an education system that needs to figure out how and when and whether to engage. That's interesting. Is there really no other example of a bottoms-up technology that has—I mean, I believe you, and I'm actually just trying to think. It makes sense, like, meaning that it's truly student-led. I don't think at this speed of change. YouTube, that was fast adoption, but it didn't really become sort of part of core infrastructure. The internet, was it bottoms-up led? No, I think it was sort of top-down technology purchases by CIOs. So I think it's the combination of sort of being grassroots-led, but also just the speed with which this washed. Yeah, and I remember when you first started at OpenAI, one of the things you said to me was like, I have never seen numbers like this, you know, and stuff that people at OpenAI take for granted. Like, you're like, my mind is like, in terms of, like, the sheer size of the impact, you know. Part of the reason I went to OpenAI, there was two things. One, towards the end of my time at Coursera, ChatGPT was launched, and our CEO, Jeff, who was super into it, was requiring all of us to chat with Chat before we Slacked him. And his requirement was that everything we said had to be smarter than what he could otherwise get out of Chat. And we would have executive team meetings, he would have sort of ChatGPT on a computer as part of the executive team, and I was like, wow. This actually means I have sort of a thought partner and someone I can reason with to sort of be my best self in the room and around the table in a way that I didn't have before. But when I got to OpenAI, you know, I had thought Coursera was the largest learning platform in the world. And what I quickly saw was that over 20% of all messages on ChatGPT were about learning. And so what I realized is that within two short years, this platform by OpenAI had become actually the largest platform for learning in the world, short of, say, YouTube or Google Search. Give us, like, an order of magnitude. You're saying almost half the messages are— Tell me the stats again, but also give me, like, the number. I'm just, like, curious. Like, that's insane. So I don't remember the exact numbers when I joined, but it was roughly— so now Chat has, I think it's something like 950 million weekly active users. But when I joined, the big aha that made me realize, wow, learning and education is, like, a big deal at this platform, is that 40% of all users were under 24, and 20% of all messages sent on the platform were connected to learning and education in some form. Yeah. So it's a youth-driven platform, or at least— it's not majority. Yeah, it was. Got it. And that there's a significant percentage of it that's being used for learning, or that that's a significant use case. So just for you specifically, like, can you remember the first time that you saw AI, or Chatty G, as you've named him, do something, you know, in education or in learning that really made you stop and think, like, oh, this is, like, this is different? So I think my aha moment connected to the power of Chat and learning was actually the summer after I left Coursera. I have a daughter, and she's dyslexic. And for years, I've seen, you know, her brother come down in the morning and read the newspaper, and I would always think to myself, like, how is this little girl going to actually access the world's knowledge? She just cannot read the paper. And we were traveling in Iceland that summer, and it was right when Chat was released in voice mode. And I remember handing it to her and said— Why don't you try this out? Maybe you can learn about the world through this. And she, I remember talking to ChatGPT, and she used, like, please and thank you. And she was like, Hey, my mom wants me to learn about current events. Can you help me? And Chat replied and said, You know, what are you interested in learning? And she started saying she was interested in astronauts and space. And the next thing I know, Chat was talking to her about the Mars rover and the history of the space program. And it was just a complete unlock. I realized, you know, I no longer had to worry about her ability to access the world's information and to sort of get that information in a way that was tailored to her interests and passions. And it blew me away. Yes. So, so it's interesting because, you know, the concept of, like, personalized learning, right? Like a tutor that knows you and understands you and can kind of, like, meet you where you are, like, which I think is, like, one of the things that people seem really hopeful about with AI. Like, that concept is not new. There's obviously, like, people that have been working on it for a long time in education. And I would say it also hasn't made as much progress as people, like, hoped it would in the last 10 years. Do you think, like, do you actually think this is the moment? Like, and if so, what can happen now that literally couldn't happen before? I think AI does have potential to one day become that personalized tutor. And just to give a little context, the reason why people are so excited about that tutor concept is there's a study by this guy Benjamin Bloom. And what he found is that when you are trying to teach a topic and you compare the results of kids who are learning in the classroom to those who are learning with a tutor, the speed at which those who are learning with a one-on-one tutor is just miles ahead. There's a huge, huge difference. But historically, you know. Obviously, there's not the opportunity in this world today for everyone to have a real-life tutor. But that's been kind of the North Star, the golden goose of education, education technology, so to speak. The honest truth is it's much harder to achieve than I think I certainly thought when I joined OpenAI, and I think it's going to be a while before we get there. But explain that, because part of what you're describing was, and I think part of even my experiences with ChatGPT and other chat interfaces is that it is like a personalized tutor. But I know that part of the challenge in education isn't always having the technology; it's actually getting people to use it, and particularly maybe also people in the classroom. Is that the barrier that you're talking about? So I think what Chat is, and all the frontier models and platforms are great at doing, is giving information in a personalized way or answering questions very specifically. But to really be a tutor, an effective sort of virtual tutor, you need to do something more. You actually need to interact with a person in a way that makes them think, makes them reflect, makes them apply information to other contexts. And if you think about, I'm sure you've had tutors for your kids, or maybe they're younger, a tutor that works with your kids, they understand what the goals are that your student wants to achieve. They understand sort of what they know and where they need to go. And it's that whole combination of interacting in a way that doesn't just give information in a personalized way, doesn't just give answers, but actually pushes someone's brain to experience friction, to reflect, to remember, to process, that allows people to learn. And that's the problem. Right now, these models are great. They can respond like a tutor. They can respond in a personalized way. But we haven't yet figured out how AI can really unlock the ability to sort of improve people's cognition, their memory, their ability to reflect. Reflect, and that's the journey. That's interesting, because, like, one of my questions was, like, what is school for then? But I think you're—well, I think you're kind of answering that, but, like, what would you say? Like, what is school for if we have access to all this stuff? So there are many theories now on why we're not yet managing to see the results of the sort of same tutoring results we saw in in-person tutoring and the AI tutors, and I sort of share with you a couple of my theories. But part of what the gap is is that ChatGPT is not actually great at motivating you, at holding you accountable, at bringing you in contact with other humans who can sort of deeply understand who you are and ignite your curiosity and desire to learn. And so part of what school does is it gives you this container, this structure for accountability, for motivation, for sort of addressing the other things that make you want to learn. It's a narrow definition of school, but I think that's part of why we see that gap. Okay, man, I have some directions we could go with this. I do want to ask you this question, though, because I feel like there's, like, a lot of, you know, chatter in the world around both, like, basically this being a tool that lets people, like, cheat or, you know, almost like it's making people dumber, you know? And it's, like, so interesting where you're like, wait a minute, we just gave people this, like, incredible piece of technology that, like, gives them all this access to information, but it's—there is a fear, and I have no idea if it's grounded or not, that it's so easy, right? It's like the calculator means you don't have to do math in your head, so we all got dumber, which I, by the way, can't do math in my head. I blame the calculator. Like, what are you seeing? How do you actually, like, think about that relative to this idea of, like, the power of the tool versus sort of, like, how it's being used? So we see Chat used in all sorts of ways. It's just a tool, and some of those ways, I think, help people become their best selves, and some of them are diminishing. But let's go back to the calculator example that you just gave. My daughter learned her math facts and is now doing long division, and she has, you know, cried tears at the kitchen table trying to learn long division. If I were to hand her that calculator as a tool while she was learning long division, there is no way she would have learned it. But the hope is that once she learns long division and has that down, later in her career in math, she could use that calculator to actually enable herself to do higher order math. And I think Chat is the same way. If you give a kid, you know, a phone and ChatGPT and they're doing their homework, and what they start to do is take a picture of their homework and say, Hey Chat, help me, and then write the answer down, it may actually feel like learning to them. It's not that they're trying to cheat. It may feel like learning, but at the end of the day, it's actually taking away that sort of friction and those tears and that struggle that are so important to learning. And I think that's what people are concerned about. The challenge we have right now is that that fear and that concern about the impact of AI on learning is also making many people try to hang back, and many schools, you know, they're not wanting to engage, teachers not wanting to engage. And my view is that, like, we as an education system need to lean in. We have an opportunity to actually help students understand what it means to learn and to guide them to use these tools in a way that are productive and helpful. And I think, you know, we didn't do that with social media since it wasn't an education tool. We're like, you know what, parents just need to deal with that. But AI is different. And so part of what I'm hoping is that even while we're on this journey to figure out what is the highest best use of AI now, we, you know, within the education system, start learning and understanding the technology and guiding students. I mean, if you believe this is a tool that's always going to be around, you don't do kids any favors if you Try to educate them in a world where you pretend it genuinely doesn't exist. I mean, I think that's, like, what I hear you saying is, like, the struggle is valuable, right? Like, that the manual learning and the reason why we all learn, even though there was spell check, we still learned how to spell, and even though there was Excel, we still learned how to do calculations and things like that. Like, there is something really valuable that happens in your brain when you don't get to use a tool. And part of what I feel like schools are struggling with is that tension, right? Like, how and where should this tool be part of education? Yeah. Totally. That is the struggle. Okay, so what do you— What's my point of view on it? Well, I'm, yeah, I'm curious, like, let's say I am one of those people, or even a parent, but certainly an educator that's like, you know, kind of like, get this out of here. I want to do things as if it doesn't exist, or ban it, or whatever. Like, what do you think they're getting wrong? So maybe I can sort of give the answer by talking about my own kid. You know, two years ago when I joined OpenAI, I was very focused on helping my son learn how to use AI in a way that was productive. So I was off in California, he was home struggling with his homework, and I said, can I actually train him to use Chat as a coach or a tutor and see how that goes? I would say fast forward two years, I'm less focused on that, and I'm more focused on how can he use AI to actually create and build. And so to tell you a quick story, you know, a week ago, my kids want to start a fidget sales business. They want to, like, make these little 3D-printed fidgets and sell them at farmers markets and online. And so we sat down and I said, let's forget the sort of learning tutoring thing for a minute, but AI has all these coding tools now and all these tools for building. Let's sit down and let's see if we can teach you to build a website. And— It was just a big unlock for him, and he started using AI, and he started coding this website, and now he's diving deep and creating things. And that experience with him of AI sort of giving him this sense of agency that he can create and build, that really shifted how I think about the value of this technology, particularly now while kind of this big learning quest is on a longer journey. Yeah, that's interesting. Well, first of all, what happened to him when he learned he had all this agency? He has just become a super motivated kid who wants to dive deep and figure out, like, what he can build and create in the world. We sent him to an AI camp because I'm now sort of using him as a guinea pig. He feels like he can build and create. And so just to take that idea a step further, to me, that's what we need to bring into education. We need to bring AI into education in a very focused way where we teach students how to build and create and solve problems and build prototypes in a way that's pretty darn focused. And I think it's going to be important. It's very different from what kids study in school today. But AI just has this huge potential to unlock a sense of agency and opportunity, and that's now where I would say I'm focused. This episode is brought to you by Wayfair. It is brought to you by Wayfair. Summer is winding down, the kids are going back to school, and I've been thinking about a few little ways to make this fall feel cozy at home. I recently did a quick browse on Wayfair and somehow ended up with a full cart. I picked up a Cuisinart electric kettle for my daily cup of tea and a full set of nonstick frying pans as a gift for my husband, who is our household chef. What I love is how easy it is to find what you're looking for using the filters and reviews on Wayfair. And if you want a shortcut to the good stuff, shop Wayfair Verified. Their product specialists hand-vet everything, so you know you're getting a quality piece. Whether your vibe is modern, coastal, or somewhere in between, it's all there. Transform your space with pieces that hold up to real life. Join Wayfair Rewards today to get 5% back on every purchase. Head to wayfair.com or shop the Wayfair app for all things home. That's W-A-Y-F-A-I-R dot com. Wayfair, every style, every home. This episode is brought to you by BILL, the intelligent finance platform that helps businesses and accounting firms scale with proven results. With AI-powered automation, BILL removes the busywork from your accounts payable workflow. They handle capturing invoices, routing approvals, and syncing with your accounting software so that your team can focus on growth instead of paperwork. BILL is so reliable, according to BILL, 98 of the top 100 accounting firms in the U.S. trust it to simplify and secure their bill payment processes. BILL's handled over a trillion dollars in secure payments and is ranked number one overall on G2's 2025 list of best accounting and finance products. So stop the guesswork and start scaling with the proven choice. Go with the company whose financial infrastructure is trusted by nearly half a million customers. Ready to talk with an expert? Visit bill.com/proven and get a $150 gift card as a thank you. That's bill.com/proven. Terms and conditions apply. See offer page for details. This episode is brought to you by Capital One. Capital One's tech team isn't just talking about multi-agentic AI. They already deployed one. It's called Chat Concierge, and it's simplifying car shopping. Using self-reflection and layered reasoning with live API checks, it doesn't just help buyers find a car they love. It helps schedule a test drive, get pre-approved for financing, and estimate trade-in value. Advanced, intuitive, and deployed. That's how they stack. That's technology at Capital One. Part of what I hear you saying, which was actually different than what I thought you were going to say, is that this idea of, like, a tutor or it being a universal teacher is feeling far away. Like, can you just, like, tell me a little more about that? Like, why is it farther away? It's farther away because I think we haven't yet figured out how to enable people to respond to an AI in the way they respond to teachers. Teachers motivate. They hold people accountable. They bring students together in community. They understand a student's context, their emotional context, their state of mind, what they know. And AI just doesn't do that. It's not there. It can't perform the human parts that make people excited to learn with a teacher or a tutor. Do you think it'll ever be able to solve that? I think that we will end up in a world where, yes, AI can tutor much more effectively than it can now, today. But I don't think that will ever be a replacement for the presence of a human actually sitting there and bringing together a group of students and motivating them to learn. Yeah. So we'll get a lot better, but it's— But it's, I don't think it'll ever be a replacement for human tutors or teachers. Interesting. So it'll be a human-robot companionship, at least for as far as you can see, which, by the way, is like a theme that's coming out. I've done a few of these, and like, it's just so interesting. All the news stories that are out there about, you know, every job in the world going away and every human being replaced, and everybody that's at the edge of AI is saying that's not what we're seeing, you know, and that's actually not the reality that seems to be coming, that this is about a change, for sure, seems like, but that it continues to be about there's a set of things that only humans can do. I agree with that. What are some of the coolest things that you've seen that make you, like, so excited about the future of this technology? So about a couple weeks ago, we pulled together a group of students. We called them sort of ChatGPT Future Students, and they were just students from all over the U.S. who were building with AI, and they were doing amazing things. One of the students I met is using models. He's trying to study solutions to sort of treatments for bipolar disorder, and he is using LLMs to look at all the available drugs on the market and figure out what drugs might be out there that aren't currently being used for mental health reasons that could now be applied to address mental health. Another student, she and her friends would sit in the cafeteria and play Wordle together, but one of their friends was blind, and so she decided that she would actually build a version of Wordle that a student was blind could play. And from there, she's launched a company to actually build educational games for students who have all sorts of disabilities, and more basic things too. You know, there's a student from Penn who joined us this summer, and he's been running an AI club at Penn, and the first thing the team did, he invited a bunch of non-technical students, and they figured out a solution to actually take the unused food from all the cafeterias and bring them to food banks. So, you know, what's really just struck me is I've been on this journey since starting in OpenAI to figure out, like, what do students actually do with this technology? Because the narrative out there is that they cheat, but what you actually see is that so many students, when they learn how to use it... are looking at sort of problems in the world that they want to solve and building solutions and not waiting for permission and not looking to hire engineers or VCs or networks or any of these things you and I had to do and probably still do. They're just building. And that's what makes me so hopeful. Yeah, that's cool. Well, like, just do me a favor and talk about that broader negative narrative, like the cheating narrative. Like, I—and I think it's both that there's like a cheating narrative and there's also sort of a, like, I don't know, like I said, it's kind of making people dumber. It's making certain things so easy. How do you actually address that in the actual work that you're doing? Like, and what do you think is actually real about that? Because, like, obviously, like, you know, anytime you're working on a huge platform, people are going to write all sorts of things about it. And so for someone, I have no idea, like, what is real. You see all these stories and I'm like, is this actually a big problem? And is it something that you all are tackling from the technology side, or do you actually think it really is not sort of like that the good outweighs the bad, I guess. I think it is a problem, and I think it needs to address from a human side, from the student side, and on the tech side. And before I go into it, the thing to know is that the people who are most concerned about brain rot and cheating and the impacts of AI on sort of cognition and brain development are students themselves. Like, students want to thrive, they want to learn, they want to get great jobs. But we've done a couple things. The first, we built out a product called Study Mode. So we, when I first joined, we sat back and we said, well, if ChatGPT is giving too many people easy answers, how do we actually build an experience in the product that would behave more like a tutor and wouldn't just give answers, but would prompt students to reflect and articulate what they know and to dive deeper? And we worked with, you know, hundreds of teachers and learning scientists and built this experience called Study Mode and saw that, you know, lots of students enter this experience and they see it as a way to have Chat help them through ideas and help them through our homework in a way that's different from just give me the answer. But the other thing we're doing is just starting to work very, very deeply with teachers, K-12 teachers, professors, because I think part of what needs to happen in society is we need to have adults guide students on how to use this technology in a way that's sort of productive and to help students actually understand themselves. You know, as you said, you were like, wait, I thought that's learning. If your daughter's getting personalized information, isn't that learning? And so sometimes people are not learning, but they don't realize. They're like, oh, I'm learning. I'm getting personalized information. So we need to educate students on what does it mean to learn? How do you learn? What is learning and what feels like learning? And bring that into the classroom as well. So it's... Yeah, I think that idea of, like, what is actual learning and what feels like learning feels like a really big tension in your work to me and in the product. Because I think you genuinely do think you're learning when you're getting all this information. But I think what you're saying is, no, you're not. Yeah. So, out of curiosity, like, I think the calculator example has always been one for me, probably because of the era I was raised in, you know, where, like, I was not allowed to use a calculator for a long time, and then I got one in calculus and algebra maybe. But by the way, I still can't do math in my head. So, do you have, like, a strong perspective about how, what role these tools end up having in the classroom? Like, how do we create good hardware you can't use them and you have to go through all that tears, to your point, all the tears and struggle, and where AI might have a role in the actual sort of, like, institutions of learning? So one thing that's very clear is that AI can be incredibly helpful to both faculty and teachers. Teachers are constantly doing a huge amount of administrative work, from grading to compiling assessments to sort of preparing new materials. And what you find is that when teachers are actively using AI, they can save, like, six, seven hours a week, spending less time in administration and actually, you know, more time in the classroom with students. That's one. I think teaching students in the classroom how to actually build and create with AI is both a skill and a mindset that's going to be very important for the future. You know, in most schools today, the idea of, like, entrepreneurship or making or coding, it's a sideshow in education. It's something that, you know, students who have access to resources or an extracurricular club, they can do. And part of that is because the tools for building have been so scarce. But now that they're not, I'm hopeful that every student in high school can go through this experience that my son went through of identifying a problem and building a prototype and picking a problem, building a prototype, and doing that again and again. So that's a very clear use. And then I think we're honestly on a journey to discover it. I'm not of the view that, you know, students should just sit there with ChatGPT open all day in the classroom and deploying AI into schools is what that means. I think if we deploy in the classroom, we need to do it in a way that's really sort of purpose-built and intended, intentional. It's so interesting because to the conversation around big narratives and all the hype and good and bad, like, you're painting actually like a pretty—narrow is not the right word—but like purpose-built, let's use your words, like purpose-built vision, which is just like, this is a tool. And like, it's like the calculator, but like the most powerful calculator in the world. Like, it is a tool to be used thoughtfully and carefully, but it is not—it's maybe an accelerant and a jetpack in some cases, but it is a tool at the end of the day, not the future of learning and taking over all schools and things like that. For sure. That's my view. And look, what I'm trying to do is sort of dual track. You know, we're working with researchers and countries all around the world who are working with us to figure out how do we advance Chat and other tools to make them as good as possible for learning. You talked about anxiety, you know, and you talked about students. And I feel like, you know, my teenage friends, let alone my young friends that are in college, are sort of like, will there be jobs? You know, and what does the future look like for me? I think partially, obviously, because of the fear-based narratives that are out there, you know. And you've got a 13-year-old. So, like, what are you—well, if you were 20 today, like, what would you be focusing on and what would you ignore? You know, I'm asked this question when I go around. I bet. You know, what are the skills that my kids should learn? What do we need to teach kids? And I always reframe the question. I think, like, a deeper, another way of asking that question is, you know, what do we want humans to be able to do in a world where intelligence is infinitely accessible? And I think, you know, what we want our students to do is to be able to sort of possess those very best human skills that we talked about when we talked about the power of teachers and the power of tutors. We want them to be able to interact and show emotional intelligence and to lead and to judge and to look at the world and perceive it. And then beyond that, you know, what I want for my kid is a sense of empowerment. And none of us know what the jobs of the future will look like, and there may be a lot of them or there may be a little of them. But what I most hope is that he will go out into the world and he'll be able to identify problems in the company that he joins, and if there's not a great company to join, then he will build his own. So I think it's that combination of the ability to go from problem to prototype and sort of solve challenges and build and create, plus sort of deep human skills that will allow you to drive powerful relationships and impact with other people. That resonates. And so you got a 13-year-old. How is this affecting your parenting? Like, what are you doing? How do you actually kind of think about where the struggle, like guide where the struggle is important and not, or are you leaving that up to the school? So the school that we are at, one thing I really admire it is very clear with students on when they want you to use technology and when they don't. Part of the challenge is that many schools aren't sort of talking about AI and tech explicitly, and so they're leaving it up to students to decide. And so, you know, he knows from his school that when he writes, and writing is hard, it's hard for students to write, that he's not allowed to use technology. And so, you know, part of what I've done is I've tried to reinforce for him when school gets hard or when homework gets hard, like that struggle is where you're learning. So don't pull out the tech and don't pull out Chat. And honestly, I may have scared the pants off of him. But, you know, I see how easy it would be when he struggles to just pull out Chat. But I think what's been powerful is that that self-awareness, in addition to the school policy, has helped him use Chat, at least for now, in ways that are pretty intentional. And that's my hope, that he can use it as a tool for a purpose when it will help him grow and learn. It's actually sort of a funny, simple rule because it is like the moment when we reach for tools, when your brain is like, This is hard, you know? And it's like, I think one of the really salient points you're making is just like that struggle is where you learn, and like by— By short-circuiting it, you, in some cases, are like almost cheating yourself, you know, from what comes out the other side of that learning. I wanted to say to you that as you've been talking, I'm like, it's like almost like we need a different word. It's like information absorption versus, like, learning. But I think it really does beg that question of, like, what is learning? Do you know what I mean? And I feel like that's what I hear you struggling with a little bit, you know? I would, you know, one way to think about it is this. I mentioned that we work very different, very closely with the country of Estonia, and I recently sat down with their minister of education. And, you know, the way she framed it is this. She's like, there's sort of different types of learning. There are sort of lower cognitive processes: learning a fact, learning a piece of information, spitting that fact back. And then there are sort of higher-order things that we want our students to be able to do specifically in an AI world. We want them to be able to reflect and create and analyze. And so, you know, part of what we're in the process of doing is figuring out what are those sort of lower orders, pushing information back and forth things that right now is a lot of what we ask kids to do in school, and if we have them do it with AI, they're not going to be able to do it. And what are those sort of deeper ways of thinking and higher-order ways of processing information that we want students to grow into? And I think over time, what we'll start doing as a society is getting more awareness of, you know, what is learning, how does the brain work, and sort of what are the types of minds and the types of behaviors we want kids to cultivate. So it is a messy space. I mean, I feel like I understand things in a different way just from this conversation in terms of, like, what the complexity of what's going on is. Because in some ways, I think, you know, I mean, if you think about what education was 40 years ago, it was like, memorize this textbook, you know what I mean? And, like, that literally was information absorption, you know, and— It sounds like part of what's going on for you, or what you're in the middle of, is like a really deep reflection on the value of structured learning and/or education, or whatever the right words are, which I think, you know, probably part of the challenge going on in the world is just that this is challenging people to have to ask that question, you know? It's challenging to both figure out how this new technology fits into our education system at the same time that we're trying to figure out what do schools need to accomplish in a world where AI exists and intelligence is infinite. And, you know, these are two very different questions, and we're having to work it all out at the same time. I love it. Leah, this, like, this was so interesting. Like, I—it's such a complicated conversation, and you're grappling with things that no one really has answers for, you know? But you're also not saying like, we have the answers. You're saying we're trying to figure it out too. Yeah. I think that that's true. It's complicated, and again, I wish more of the industry and the news media would engage directly with the students and understand, like, what is it that they see unlocking for them, and what opportunities is it creating? Because I think if we get clearer on, you know, what problems is it solving, what does it allow them to create and build, we will move away from, like, the fear and the worry and the sense that AI is going to, you know, collapse all these things and really dive deep into this problem of, like, what can AI open up for us? I love it. Leah, thank you so much. I really appreciate this, and I feel like I'm going to be thinking about this as I think about my kids too. I like your point about the struggle. Like, the struggle's the point. How do you help them learn that? It's not easy when you see those tears, but it is important. But Molly, thank you so much for inviting me and I would say bringing the conversation down to earth and asking hard questions. I appreciate the way you engage in the world. Okay, that was such an interesting conversation with Leah. I honestly, it didn't go the direction I thought it was going to. I feel like I learned a lot from that, but I guess just because of maybe the narratives in the press, and I thought we were going to hear a lot more about, like, AI replacing everything in education. And that's really not what I heard. What I heard was, like, a deep conversation and struggle to some extent around, like, what is learning and what is school for? Which I feel like talking to Leah just really made it clear that just having access to information or a sort of infinitely patient friend that has a lot of the world's knowledge inside of it doesn't, in and of itself, create learning. You know, anybody that's been through a deep learning experience knows that feeling of, like, it getting hard and making it to the other side of that, and that embedding itself in your brain in a different way. But the most important thing, you know, was just a really direct message that this is a tool. You know, so many of the conversations happening out in the world are based in this fear-based language that, like, AI is replacing everything and AI is going to be everything to everyone and all jobs are going away. I don't know, there's going to be no use for humans anymore. And, you know, a lot of what I heard in this conversation was there are some things that AI is not going to be able to do anytime soon that humans are really important for, and— I guess the other side of that that I heard was like, it's a tool that's not going away. It's not only a tool that's not going away, but every single kid that has access to it is using it and wants to use it and wants to help it power their learning or power their lives. But they're struggling with how to use that tool. You know, for all of us, whether you're a parent or just someone that's thinking through this yourself, the question is, what should our relationship with it be, right? What should our kids' relationship with it be? How do we want this tool to be a part of our lives and a part of our learning? I don't think we have the answer, but it sure has left me with a lot to think about. I hope you found that as interesting as I did. All right, I'll talk to you all next week. Work Life is a production of TED and Pushkin Industries. This episode was produced by Isaac Carter and Leah Rose. Banban Chang is our story editor. Mixing by Hansdale Shee. TED's executive producer is Daniela Ballarezo. Constanza Gallardo is the executive producer for Pushkin. Special thanks to Roxanne Highlash, Valentina Bohanini, Lainey Lott, Tansika Sungmanivong, and Ashley Murphy. If you like the show and want more, come join the discussion on my Substack, Lessons. I'm Molly Graham. Thanks for listening. Hey, Yaties, this is Nick and Jack from the Best One Yet podcast. Now, the last company we worked at, they used Paylocity and everything just worked. It wasn't until launching our own media business, this show, that we realized how rare that is. Because Paylocity is one delicious burrito of operational needs. They roll up HR, finance, and IT seamlessly into one delicious bite. When everything wraps together like that all at once, your workforce, your tech stack, your business, you don't need more tools. You don't even need Solandra. You need one solution. And that is why Paylocity built a single platform to connect HR, finance, and IT with AI-driven insights and automated workflows that simplify the complex and power what's next. Or as we call it, a delicious operational burrito. Yes, we do. Experience a one place for all your HCM needs, besties. So start now at paylocity.com/one. Paylocity.com/one. With no fees or minimums on checking accounts, it's no wonder the Capital One bank guy is so passionate about banking with Capital One. If he were here, he wouldn't just tell you about no fees or minimums. He'd also talk about how most Capital One Cafes are open seven days a week to assist with your banking needs. Yep, even on weekends. It's pretty much all he talks about. In a good way. What's in your wallet? Terms apply. See capitalone.com slash bank. Capital One, N.A., member FDIC. Hey, it's Paige DeSorbo from Giggly Squad, and this episode is supported by Rothy's, the brand famous for chic, comfortable flats. My listeners know I love to start an outfit with the shoes, and there's nothing better than a great flat in the fall. They're timeless, they make everything look more polished, and Rothy's iconic pointed flat, The Point, has been reimagined to be better than ever. It now has a cushioned, bouncy insole with much more arch support, so it's comfortable enough to wear all day. And The Point comes in so many gorgeous colors, patterns, and textures. Discover The Point at rothys.com. That's R-O-T-H-Y-S dot com.