I have been fortunate to grow up with Google for most of my life, and so it always seemed very obvious to me that you can really teach yourself anything.
Super interesting article, Job. I'd love to know your thoughts on EdTech, books vs screen learning, and how we should view the "world's greatest teacher" when it comes to our children....
I think we live in the golden age for edtech ..potentially. I haven't seen it realized yet, but it's now so much easier to fill the curriculum requirement - so people can focus more on the greater journey, creating great feedback mechanisms, etc.
That said, I do think it's crucial our kids learn to focus in the absence of perfected attention-grabbing mechanisms. My kids aren't allowed any screens during the week, and limited in the weekend.
Reading is one of the greatest pleasures at the lowest cost, but it's a skill you have to develop, a love that you have to let grow. I'm not confident there are other ways than to just read to get there.
And so would you relax your screen free weekday rule if your children were given screen-based homework? Here in the UK, homework is often online 'quizlets' and gamified repetition apps...
I’d not stand in the way for their education. But only for the homework. Not then other screen activities.
I don’t think screens are inherently bad. Just that they have lots of bad content. Whether the homework is bad or not.. depends on what the school decides I suppose!
I watched it happen in real time. Between 2005 and 2008, something shifted in classrooms that has never shifted back.
Facebook arrived. Then video social media. Then the dopamine-optimized content machine that now defines how an entire generation processes information.
Schools have been fighting a war for attention ever since. And they're losing badly.
The evidence is everywhere. 77% of U.S. public schools now prohibit cellphone use during class. Between 2024 and 2025, almost two-thirds of U.S. states adopted policies to restrict phones during the school day. The only way schools can get students to pay attention is to physically remove the competition from the room.
Two generations ago, schools were a nice-to-have, not a necessity. You could wash dishes, start a restaurant, and build a comfortable life without formal education. You could farm. You could work with your hands. You could create financial security through trade skills and entrepreneurship without ever stepping into a university.
But by the late 20th century, something fundamental changed. Schools and universities exploded in popularity because knowledge work became the primary path to a comfortable life. For Millennials and the generations that followed, education became mandatory - the clearest and often only route to economic security. That's still largely true today.
Schools became central to society because they delivered on a clear promise: acquire knowledge and credentials, land a knowledge work job, live comfortably. And for decades, schools were successful because they prepared students for exactly that kind of work.
But here's what's breaking: knowledge work itself has fundamentally changed, especially for Gen Z entering the workforce now.
Modern knowledge work is less about the knowledge part and more about long-term horizon goals, complex problem-solving, and - crucially - the ability to focus without being distracted by procrastination.
The classroom model still looks like it did before the internet. Stand at the front. Deliver information. Test recall. Repeat.
But we live in a world where knowledge is commoditized. You can learn anything from anyone at any time for free. The entire educational model is built around providing knowledge in an era where knowledge is the least scarce resource we have.
Schools are still optimized for knowledge transfer and recall - the skills that mattered when education became mandatory. Meanwhile, the skills that actually matter now go undertaught: sustaining focus on long-horizon tasks, tech savviness, strategic problem-solving, and soft skills. We're building students for the last century's version of knowledge work while the current version demands something completely different.
Students know this. They feel it.
That's why they reach for their phones.
Here's the uncomfortable truth I don't hear people saying out loud: AI has created a credibility crisis in education.
Why would a student trust a teacher who knows less than the device in their pocket?
AI knows more than teachers on basically every topic taught at undergraduate level or lower. It can explain concepts countless times at any level. It answers every question without splitting attention across a classroom of 30 students. It never gets tired, never gets frustrated, and it's available at 3 AM before an exam.
The data backs this up. A 2025 randomized controlled trial found that students using AI tutors performed substantially better than those in traditional active learning classrooms. We're talking effect sizes of 0.73 to 1.3 standard deviations - one of the largest effect sizes ever recorded in education research.
Students who actively engaged with AI tutors achieved up to 15 percentile points higher grades compared to parallel courses without AI support.
The machines are already better at the core function we built schools to perform.
What AI Already Does Better
AI tutors offer something no human teacher has ever been able to deliver at scale: true personalization.
They provide adaptive learning pathways. Real-time feedback. Tailored content that adjusts to each student's grasp of concepts. Unlimited patience and availability. No judgment when you ask the "stupid question" for the fifth time.
A 2025 global systematic review found AI consistently outperforms traditional methods when measuring learning effectiveness. The evidence base is no longer debatable.
Right now, AI tutors have two critical gaps that prevent them from completely replacing traditional education.
First: persistent memory.
Current AI tutors forget. They can't track a student's journey across sessions, across a semester, across an entire academic career. They can't build on what you learned last week or identify patterns in how you think.
When AI gets persistent memory, it unlocks personalization beyond a single session. Personalization over a semester. Over a degree. Over your entire learning life.
Second: curriculum coherence.
AI tutors today can't follow a curriculum consistently. They forget where they are. They skip content. They don't present information in a logical scaffolding order.
A curriculum isn't just a list of topics. It's a carefully sequenced presentation of information that builds understanding step by step. AI tutors can't maintain that coherence yet.
But they will.
And when they do - when AI can both remember your entire learning journey AND maintain logical scaffolding while adapting in real-time - that's when the tsunami hits.
By 2027, we'll have AI tutors that are personalized, adaptive, guided, low-cost, and always available. They'll remember everything about how you learn. They'll follow coherent curricula while diagnosing emerging gaps and addressing them in real-time.
The same thing that happened to chess players after Deep Blue. The same thing that happened to local theaters after Hollywood. The same thing that's happening right now to software engineers.
Super interesting article, Job. I'd love to know your thoughts on EdTech, books vs screen learning, and how we should view the "world's greatest teacher" when it comes to our children....
I think we live in the golden age for edtech ..potentially. I haven't seen it realized yet, but it's now so much easier to fill the curriculum requirement - so people can focus more on the greater journey, creating great feedback mechanisms, etc.
That said, I do think it's crucial our kids learn to focus in the absence of perfected attention-grabbing mechanisms. My kids aren't allowed any screens during the week, and limited in the weekend.
Reading is one of the greatest pleasures at the lowest cost, but it's a skill you have to develop, a love that you have to let grow. I'm not confident there are other ways than to just read to get there.
And so would you relax your screen free weekday rule if your children were given screen-based homework? Here in the UK, homework is often online 'quizlets' and gamified repetition apps...
I’d not stand in the way for their education. But only for the homework. Not then other screen activities.
I don’t think screens are inherently bad. Just that they have lots of bad content. Whether the homework is bad or not.. depends on what the school decides I suppose!
I watched it happen in real time. Between 2005 and 2008, something shifted in classrooms that has never shifted back.
Facebook arrived. Then video social media. Then the dopamine-optimized content machine that now defines how an entire generation processes information.
Schools have been fighting a war for attention ever since. And they're losing badly.
The evidence is everywhere. 77% of U.S. public schools now prohibit cellphone use during class. Between 2024 and 2025, almost two-thirds of U.S. states adopted policies to restrict phones during the school day. The only way schools can get students to pay attention is to physically remove the competition from the room.
Two generations ago, schools were a nice-to-have, not a necessity. You could wash dishes, start a restaurant, and build a comfortable life without formal education. You could farm. You could work with your hands. You could create financial security through trade skills and entrepreneurship without ever stepping into a university.
But by the late 20th century, something fundamental changed. Schools and universities exploded in popularity because knowledge work became the primary path to a comfortable life. For Millennials and the generations that followed, education became mandatory - the clearest and often only route to economic security. That's still largely true today.
Schools became central to society because they delivered on a clear promise: acquire knowledge and credentials, land a knowledge work job, live comfortably. And for decades, schools were successful because they prepared students for exactly that kind of work.
But here's what's breaking: knowledge work itself has fundamentally changed, especially for Gen Z entering the workforce now.
Modern knowledge work is less about the knowledge part and more about long-term horizon goals, complex problem-solving, and - crucially - the ability to focus without being distracted by procrastination.
The classroom model still looks like it did before the internet. Stand at the front. Deliver information. Test recall. Repeat.
But we live in a world where knowledge is commoditized. You can learn anything from anyone at any time for free. The entire educational model is built around providing knowledge in an era where knowledge is the least scarce resource we have.
Schools are still optimized for knowledge transfer and recall - the skills that mattered when education became mandatory. Meanwhile, the skills that actually matter now go undertaught: sustaining focus on long-horizon tasks, tech savviness, strategic problem-solving, and soft skills. We're building students for the last century's version of knowledge work while the current version demands something completely different.
Students know this. They feel it.
That's why they reach for their phones.
Here's the uncomfortable truth I don't hear people saying out loud: AI has created a credibility crisis in education.
Why would a student trust a teacher who knows less than the device in their pocket?
AI knows more than teachers on basically every topic taught at undergraduate level or lower. It can explain concepts countless times at any level. It answers every question without splitting attention across a classroom of 30 students. It never gets tired, never gets frustrated, and it's available at 3 AM before an exam.
The data backs this up. A 2025 randomized controlled trial found that students using AI tutors performed substantially better than those in traditional active learning classrooms. We're talking effect sizes of 0.73 to 1.3 standard deviations - one of the largest effect sizes ever recorded in education research.
Students who actively engaged with AI tutors achieved up to 15 percentile points higher grades compared to parallel courses without AI support.
The machines are already better at the core function we built schools to perform.
What AI Already Does Better
AI tutors offer something no human teacher has ever been able to deliver at scale: true personalization.
They provide adaptive learning pathways. Real-time feedback. Tailored content that adjusts to each student's grasp of concepts. Unlimited patience and availability. No judgment when you ask the "stupid question" for the fifth time.
A 2025 global systematic review found AI consistently outperforms traditional methods when measuring learning effectiveness. The evidence base is no longer debatable.
Right now, AI tutors have two critical gaps that prevent them from completely replacing traditional education.
First: persistent memory.
Current AI tutors forget. They can't track a student's journey across sessions, across a semester, across an entire academic career. They can't build on what you learned last week or identify patterns in how you think.
When AI gets persistent memory, it unlocks personalization beyond a single session. Personalization over a semester. Over a degree. Over your entire learning life.
Second: curriculum coherence.
AI tutors today can't follow a curriculum consistently. They forget where they are. They skip content. They don't present information in a logical scaffolding order.
A curriculum isn't just a list of topics. It's a carefully sequenced presentation of information that builds understanding step by step. AI tutors can't maintain that coherence yet.
But they will.
And when they do - when AI can both remember your entire learning journey AND maintain logical scaffolding while adapting in real-time - that's when the tsunami hits.
By 2027, we'll have AI tutors that are personalized, adaptive, guided, low-cost, and always available. They'll remember everything about how you learn. They'll follow coherent curricula while diagnosing emerging gaps and addressing them in real-time.
The same thing that happened to chess players after Deep Blue. The same thing that happened to local theaters after Hollywood. The same thing that's happening right now to software engineers.