Estonia has pioneered a nationwide experiment in high schools to integrate AI into education, launching the "AI Leap" program that provides every 10th and 11th grader with access to ChatGPT. The initiative includes a Socratic Chatbot designed to foster critical thinking by asking guiding questions rather than delivering answers directly. While the government and educators aim to preserve cognitive development and active learning, students report frustration with the tool—particularly its tendency to switch between languages and its persistent questioning. Usage remains low, with only about 65% of students activating accounts and a minority using the tool regularly. Students express varied opinions: some use AI to overcome weaknesses in subjects like math, others avoid it due to fears of dependency or environmental costs. A 12th grader, Cordelia Violet-Pap, strongly opposes AI, warning that students will use it to bypass effort and that it risks eroding educational value. The experiment highlights a central dilemma in AI education: how to harness technology’s potential while protecting essential learning skills. Ultimately, Estonia’s experience offers a cautionary and instructive model for schools worldwide, suggesting that AI integration must be carefully designed, student-centered, and ethically grounded to avoid undermining the core purpose of education.
A few months ago, my colleague Sam Shekner flew to Estonia in northern Europe. He was there to visit high schools. A Spanish teacher let him sit in on the class. He had set up the class in the computer lab and he introduced me to the class and said I'd be observing. The whole text that you have there just copy paste that into chat and you all I have also a text. And, you know, basically he showed them some sort of game-like activities that he'd created with ChatGPT for them to practice their Spanish. And then give ChatGPT the prompt that you have there. Yes, my dear, it's the same one, yes, so you should have. Sam, how did you find yourself in a classroom in Estonia? Well, they are doing something that I thought was unusual, interesting, and a little bit scary. They are giving every high school student in the country, or at least to start with every 10th and 11th grader their own ChatGPT account. Why? Well, yeah, that's the funny part. They're doing it because they're worried about the impact that AI could have on learning. Estonia started thinking about the project in 2024. After one survey found that 90% of the country's high school students were already using generative AI for learning. They figured that not doing anything was the worst choice of all. And so instead they're launching a national experiment to figure out what the best thing is to do. And I wanted to figure out if there's a way to, I guess at a basic level, if there's a way to keep learning in an age when machines can do your thinking for you. If you don't have to worry about the grammar, it just, if the chat can tell you, if you have made a mistake, and you can correct it afterwards. Why does what's happening in Estonian matter to people who are not living in Estonia? Estonia, right. It's a small country, it's bordering Russia, it's different language, it feels pretty far away. But this is really the first big deployment in high schools of ChatGPT, the entire idea of school and how we encourage learning and what even we want people to learn is that's take here. And so it's a glimpse, maybe, of what could happen in a school district near you. Welcome to the journal, our show about money, business, and power. I'm Ryan Knutson, it's Thursday, September 24th. Coming up on the show, Inside Estonia's high school experiment with AI. Imagine setting your makeup, then forgetting it's even theirs. Meet new grippy setting mist from Mavily New York. Gel to mist technology locks in your look for up to 24 hours, with flexible all-day comfy grip, no tightness, no stickiness, no residue, just plump, dewy, hydrated skin that still feels like your skin. Try new grippy setting mist from Mavily New York. Estonia's decision to bring AI into classrooms is not the country's first rodeo with new technology. Estonia, since it gained independence from the Soviet Union in the 90s, was early to put the internet in classrooms. They had something called the Tiger Leap. Estonia's Tiger Leap foundations launched a program introducing IT Science and Computer Programming at schools from the first grade. During the Tiger Leap, new shows in Estonia showed kids rushing toward computers. Its designers point out that internet and modern technologies have become an integral part of daily life. They're extremely proud of this. This is part of the country's kind of founding legacy. They see themselves as a digital country. And so, when the AI revolution comes about, the natural instinct is, okay, we're going to have to grab this bull by the horns. This is what Estonia does. So last year, Estonia decided to launch a new experiment. They're calling it the AI leap. The country's partnered with OpenAI and recently with Google to give tens of thousands of high schoolers access to free chatbots. One school that's part of AI leap is located in Tallinn, Estonia's capital. During his visit, Sam met with the school principal. Indrek Lilimagi, he is the principal of Tallinn Pellgillina State High School, or Perg, as it's known. And, you know, Indrek himself uses AI tools. He, you know, tells me he does some vibe coding and, you know, the school ethos that he set up is very human and, in fact, in the school, they don't oblige any student to use AI if they don't want to. AI leap, the program has provided meaningful, high quality materials, resources, workshops for teachers, school leaders. I think it's going well. I mean, I guess one argument would be that, effectively, you're saying, use AI and that it encourages more than my other ones. Yeah, but the thing is that they use it anyway, they already use it. So, yeah, so I don't believe that AI leap, like, pushes them to use it more, yeah. They use it. Students might already be using AI, but Indrek told Sam that he's worried about students using the tools on their own and how it could affect their ability to learn. If you imagine that school is a gym for the brain and you go to the gym, it would be kind of pointless to ask your personal trainer to lift your weights for you, right? I mean, the personal trainer there is to tell you what weights to lift in what order, but if you don't lift the weights, then obviously you're not going to get stronger. Yeah, and your personal trainer just gets shredded. Yeah, well, I mean, as we know, these AI tools are, you know, when it comes to answering questions, they are shredded, right? They could do the work. They can do your high school work for you. AI can take a whole bunch of texts and can synthesize them for you and, you know, kind of reading something long and figuring out what's important and what it's connected to. That's thinking too. And those are important skills that you learn in school and, you know, how much you want the machine to do that versus how much you want to do it yourself or you want your kids to do it in school is a really important question. Back in the day, the way teachers got students to exercise their minds was by having them do stuff like answer questions or write essays. Teachers would then judge the output, the finished product, but the effort that the student put in to get that output was the point. That kind of learning changed with AI. AI basically breaks the connection between effort and output. AI now makes it so that the output, the performance is not correlated anymore with the effort that you do. And so you have to rethink your whole pedagogy because if the point is the effort, then the grade or the output kind of don't really matter anymore. So how do you lean into AI? When AI breaks the learning process, Estonia's solution, in part, was to come up with a version of AI that didn't just give you all the answers. They call it the Socratic Chatbot, Socrates, the Greek philosopher, was known for not stating what he thought, but instead eliciting answers from his students by asking questions. That's sort of the the coinage of the Socratic method. Under the AI leap, Estonia took the Socratic method and applied it to the school's version of ChatGPT. Effectively, it's a tool that won't answer your homework questions for you, but instead helps you scaffold a way to approach the answer, helps you plan what you're going to do to do the assignment. And they're going to try to figure out a way to use it in a way that promotes learning rather than undermines it. So rather than being just an answer machine, which it is now from most users that want to use it, more of like a tutor almost. Exactly. This Socratic Chatbot, in some ways, mirrors the philosophy that Indrick and Perger are espousing, which is much more of a processed, based approach to learning than an outcome-based approach to learning. It's not hard to imagine some version of the future where the Socratic Chatbot does all the teaching, or each student has their own personal AI teacher. But that is not what Indrick Lillamagi's school was pushing for.
wants his teachers to play a more active role. One teacher that Sam met at the school is Agna Cossk, the head languages teacher. Sam watched his Agna try to get students in an intermediate English class to use this
ocratic chatbot during an exercise. Agna Cossk was looking at trying to get the students to, at once, practice their English by chatting with a chatbot, but doing something that would be of interest to them, which is researching possible study abroad programs in English. Now how would you do that, how would you prompt it to do that, and they talked about prompting for a while? If I say please get a career for the studies advice from me, how would you do this? And then she shared her prompt of how she was going to do it? How did I make this prompt? This is an important point to remember. Of course I asked you to make the prompt, I gave it a little bit of information. So I have this high school class, I gave it the task that we are going to do at the end. Sam looked over one student's shoulder and she tried out the prompt. By the way kids in a stony refer to the chatbot as he. At the end of the assignment, Agna asked her students to think beyond the answers that the chatbot gave them, to take the next step and think for themselves. What was obvious to Sam was that Estonian educators have put a lot of thought and care into how they bring AI in the classroom, and the students, surely they are loving and education-oriented tool designed by grownups, right? I'm curious how do you use AI? No, we don't. You don't? No. What the students think of AI leap is next. Meet new groupie setting mist from Mabelie, New York. Gel to mist technology locks in your look for up to 24 hours, with flexible all-day comfy grip. No tightness, no stickiness, no residue. Just plump, dewy hydrated skin that still feels like your skin. Try new groupie setting mist from Mabelie, New York. Maybe it's Mabelie. So let's talk about the students. How are the students in Estonia feeling about this experiment? Well, they actually did a survey of students, and I talked to a bunch as well, and one of the first and biggest reactions they have to the Socratic chatbot is that it is super annoying. They're complaints start with basic things. In Agnacosks English class, for example, the chatbot kept switching between Estonian and English. It said that I will talk to you in Estonia to support you better. Even though the prompt said please you speak to English, it's so smart. Kids started to see that it would switch back into Estonian and they had to kind of repeatedly tell the chatbot. No, this is an English class. My teacher says we should be talking in English, please stay in English. But the bigger nuisance was the Socratic element of the chatbot, where it asks more questions instead of just giving the answer. I spoke to some kids who are basically like I'm so over it, I get it, you know, AI is going to make my brain rot, but like it can't, you know, can a please stop asking me questions. Like, I don't know, what did it tell you when you wanted to make porridge? Yes, I asked him that. Oh, how I can make this porridge and the answer that he gave, what experience you have with porridge? And they haven't used it after? Yeah. I used the normal changeability of this study, even for school. Yes? Yeah. Another student uses chatbot sometimes, but actually avoids AI when it's a subject he really cares about. Mathematics and stuff, AI, a lot of times wrong, but it wrong. So it's a way, in a way, you're deciding what you think is where you want to practice and where you want to get stronger. When I feel like I'm stronger something, I use AI when I feel like I'm weak or something I especially learn it, and make sense. And then there are the kids who don't want to use AI at all. I really didn't. I think it was a waste of my time. I don't use it because I'm afraid I'll just pick up lazy. It's easy to think that you don't need to do that job, but then who does that job for you and for what cost? There are some students I spoke with who don't want to use AI tools because they are concerned about the environmental impact of data centers and, you know, water consumption and energy consumption. I just find it really wasteful, like, in the terms of water, of course. We were told in, like, third grade that we already don't have much drinking water on earth, and now they use it to pull down the servers that's like, it's terrible. At the end of last school year, about 65% of Estonian students with access to the school Socratic version of Chatchee PT had activated their account, but only a minority of students use it regularly. This year Estonia is opening up the AI assistant to 12 graders and vocational school students too. At the school in Tallinn, Sam spoke to the student body president, a 12th grader named Cordelia Violet-Pap. She is really adamantly opposed to AI. I'm really, really strongly against it, overall, just specifically the language models and generate the AI. She just thinks it's new, it's not studied yet, and she thinks that there's so much that could go wrong with it, you know, teachers don't necessarily know how to use it effectively, she said, and students don't know anything either, she said. Cordelia also said that all the thought and care that went into the school's Socratic Chatbot, does it really fix the bigger problems with AI and learning? Just the fact that we have been given this does not stop students from using normal Chatchee PT and just getting the answer anyway, which is what they are going to do because it's easy, and humans naturally want to do things that are easy, you know, school is already seen as something that's just mostly meaningless and too hard and too difficult, then why do we have to do all this stuff basically? So if such a thing is given to us that makes us able to shortcut through all of this annoying bad stuff that no one wants to do, I mean the students specifically, then people are going to use it. Cordelia has since graduated, but she told Sam that earlier this year she stood up in front of the school at a talent show and read a poem she wrote about AI. Let me see, maybe this one, this is a bit interesting because it uses words in Estonian that have been created recently. You know, one of the paragraphs that she read to me in Estonian, she said was, you know, about all of the new vocabulary that AI was kind of unleashing on all of us. I don't even know how to translate them, but maybe it's just interesting to hear for you. The poem is about Estonian words themselves, so Cordelia says the English translation sort of misses the point, but here it is. She writes, like, why must there be a term for water bankruptcy and digital trash, smart sive, AI blunder, brain rot, these are so ugly and for what? We don't need a way to express yet another world. What if the answer to this big AI experiment in the classroom turns out to be that this is a bad idea and it's not working.
at harm's education. - You know, I think that's part of the entire landscape within which the AI at Leapis is trying to navigate. You know, how can we make this a positive thing as opposed to worrying about these downsides and just saying we're just gonna say no and put our heads in the sand. And, you know, it points up a really deep tension when it comes to AI and education. Because like most of us actually, there are two minds about it. They're scared of it and they also see great opportunity and they don't wanna get left behind. - What do you think we'll learn from Estonia? - We're gonna just have to see how the experiment turns out, you know, and what the educational results are in Estonia. Are we gonna see kids doing better? Are we gonna see that this is, you know, helping educational outcomes? Because if Estonia can shard a path for this to improve educational outcomes, then that really could, you know, make an impact on schools, you know, all the way from Boise to Botswana. - That's all for today. Thursday, September 24th. The journal is a co-production of Spotify and the Wall Street Journal. If you like our show, follow us on Spotify or wherever you get your podcasts. We're out every week, day after, noon. Thanks for listening. See you tomorrow. - Imagine setting your makeup, then forgetting it's even theirs. Meet new groupie setting mist from Maybelline, New York. Just clump to a hydrated skin that still feels like your skin. Try new groupie setting mist from Maybelline, New York. - Maybe it's Maybelline.
Podcast Summary
Key Points:
Estonia has launched a national AI experiment in high schools, giving every 10th and 11th grader a ChatGPT account to address student use of generative AI for learning.
The initiative, known as the "AI Leap," includes a Socratic Chatbot designed to ask guiding questions instead of providing direct answers, promoting critical thinking over passive learning.
Despite the intention, students find the Socratic Chatbot frustrating due to language switching and excessive questioning, leading to low regular usage.
Students have diverse views on AI
A 12th grader, Cordelia Violet-Pap, strongly opposes AI, arguing that students will use it for shortcuts and that it undermines meaningful education.
The Estonian model reflects a broader tension in education
Only about 65% of students activated their accounts, and usage remains limited, indicating that AI adoption is not automatic or universally accepted.
The experiment serves as a global case study for how AI integration in classrooms can either support or undermine core learning principles.
Summary:
Estonia has pioneered a nationwide experiment in high schools to integrate AI into education, launching the "AI Leap" program that provides every 10th and 11th grader with access to ChatGPT. The initiative includes a Socratic Chatbot designed to foster critical thinking by asking guiding questions rather than delivering answers directly. While the government and educators aim to preserve cognitive development and active learning, students report frustration with the tool—particularly its tendency to switch between languages and its persistent questioning.
Usage remains low, with only about 65% of students activating accounts and a minority using the tool regularly. Students express varied opinions: some use AI to overcome weaknesses in subjects like math, others avoid it due to fears of dependency or environmental costs. A 12th grader, Cordelia Violet-Pap, strongly opposes AI, warning that students will use it to bypass effort and that it risks eroding educational value.
The experiment highlights a central dilemma in AI education: how to harness technology’s potential while protecting essential learning skills. Ultimately, Estonia’s experience offers a cautionary and instructive model for schools worldwide, suggesting that AI integration must be carefully designed, student-centered, and ethically grounded to avoid undermining the core purpose of education.
FAQs
Estonia introduced AI tools because 90% of high school students were already using generative AI for learning. To address concerns about AI’s impact, they launched a national experiment to understand how AI can support, rather than replace, education.
The Socratic Chatbot uses a question-based approach inspired by the ancient philosopher Socrates. Instead of giving direct answers, it helps students plan and think through problems, promoting critical thinking and effort over instant results.
Teachers use the chatbot to guide students through research and problem-solving—like exploring study-abroad programs in English—while encouraging students to think critically and reflect beyond the chatbot’s responses.
Students find the chatbot annoying due to language switching and excessive questioning. Many feel it disrupts learning and are reluctant to use it, especially in subjects they already understand well.
About 65% of students with access activated their accounts, but only a minority use the tools regularly. Many avoid AI in subjects like math where it performs poorly, or avoid it altogether due to concerns about effort, ethics, or environmental costs.
As a digital pioneer, Estonia’s approach offers a real-world model for how schools can integrate AI thoughtfully. Its lessons could influence education systems globally, from the U.S. to Africa.
Chat with AI
Loading...
Pro features
Go deeper with this episode
Unlock creator-grade tools that turn any transcript into show notes and subtitle files.