How to Learn in the Age of AI

Mike Lamb
13 minutes ago
9 min read

The 20th century brought an education revolution. Could the 21st century be on the brink of another?

In 1900, only one in three children went to school. By the turn of the millennium, it was four in five – an astonishing leap. The mission to widen access to education was a world-changing success.

But in 2025, it’s starting to run out of steam.

Schoolchildren have all the world’s information at their fingertips – and AI assistants to help do their homework – yet classrooms, curriculums and exams haven’t changed all that much in a century.

In higher education, tuition fees keep rising, universities are under financial strain, and the number of young people enrolling has started to fall. Some big employers are now even looking past university education altogether.

So is this the end of education as we know it? Or could it be the disruptive first act of a new golden age of learning?

Watch my short explainer video to learn more.

Out with the old school

If we invented schools today, would they look anything like they do now?

That’s the big question driving economist Steven Levitt – author of Freakonomics and founder of the Levitt Lab, a new school in Arizona rethinking what learning could be.

“In a world where you don’t know what you’ll be doing five years from now,” he told Google’s Chief Technologist Ben Gomes, “cramming your head full of facts and formulas isn’t the answer to managing the future. It’s self-evident.”

For too long, says Levitt, we’ve just been layering tiny tweaks on top of an education system developed for the 19th century – and lost sight of its purpose. “The objective should be to create a generation of people who are well-adjusted and curious and excited about the world.”

So what’s stopping us? Levitt blames grades. Exams and evaluations – a pretty fundamental part of our education system – are, he argues, at odds with the ambition to give everyone a good education.

He admits that, in his teaching role at the University of Chicago, he’s been part of the problem:

I need to find a way to give some kids As and some kids Bs. And the easiest way to do that is to teach them things that don’t matter in ways that are somewhat hard and overcomplicated so that I have a way to separate them out. And it’s terrible and I’m embarrassed about it.

Levitt is far from alone in this view. Others argue that the entire education system is built on a model that no longer makes sense in the 21st century: the scarcity model.

Could universities go the way of Blockbuster?

Before the 20th century, education was so scarce that you were unlikely to go to school unless you were a wealthy, white man. Today, it is thankfully far more accessible.

But in The Abundant University, Carnegie Mellon professor Michael D. Smith argues that higher education still runs on a “factory system” that requires scarcity to operate: a limited number of students, a finite amount of teaching capacity and a rationing of credentials.

You can sort of see why – we have a tendency to place more value on things that are in shorter supply. But as Smith puts it:

Our current system of higher education, despite its good intentions, is financially and morally unsustainable.

The scarcity model made sense when information was limited and expensive. It doesn’t in a digital world where information is virtually free.

In the film Good Will Hunting, Matt Damon’s self-taught prodigy skewers a Harvard poser trying to humiliate his friend in a bar: “You dropped 150 grand on [an] education you could’ve got for $1.50 in late charges at the public library.”

In the late 1990s, you might have watched that scene on a rented VHS from Blockbuster – as long as someone else hadn’t taken it out already.

Today, the digital revolution has swept away scarcity in entertainment and completely turned the film and music industries upside down.

What’s stopping it from doing the same for education?

Too much information

Herbert Simon, the Nobel Prize-winning economist and cognitive scientist, warned decades ago that “a wealth of information creates a poverty of attention.”

We now have access to a staggering wealth of information, yet many of us feel more overwhelmed than enlightened. And while Wikipedia contains more knowledge than could ever be held in the world’s best and biggest libraries, there’s an important distinction to be made: education is about more than just information.

MOOCs – Massive Open Online Courses – were meant to be the solution. They launched with a wave of hype and the lofty promise that education could become truly universal for the very first time. Time Magazine even called 2012 the “year of the MOOC”.

Within months, hundreds of thousands were signing up to courses and taking up the opportunity to study at Harvard and Stanford – for free or minimal cost, and from anywhere in the world.

Yellow "MOOCs" text on a black grid background. Below it, "Massive Open Online Courses" in white font.

But it didn’t quite live up to the hype. Yes, it scaled up access to elite courses previously only accessible to a privileged few. But less than 10% of students actually completed the courses. And those who did were typically the ones who were already the most academically qualified.

Today, the term MOOCs has largely been retired – probably for the best – but for the eager learner, there are countless open online courses to choose from.

One reason MOOCs failed to revolutionise education in the way some thought they might is simple: lecture videos and reading materials can scale to millions, but the most valuable resource in education can’t – individual support from a good teacher.

That, however, may be changing.

A personal tutor for everyone?

In 1984, educational psychologist Benjamin Bloom discovered something extraordinary. When students received one-to-one tuition using a “mastery learning” approach (more on that shortly), they performed significantly better than students educated in a normal classroom environment. The difference in results wasn’t minor. It was enough to turn a C-grade learner into an A-grader.

Bloom had first proposed “mastery learning” in the 1960s. The premise, in simple terms, was that students should master the basics before moving on to learn more about a subject. After all, what chance do you stand with, say, calculus if you haven’t nailed basic arithmetic?

But there was a catch. One-to-one mastery learning worked miracles, but it couldn’t be scaled. So we were stuck with the classroom model.

Now we may finally have a solution – AI.

Khan Academy is one of the leading providers of free online lessons. Its founder, Sal Khan, believes that artificial intelligence could “spark the greatest positive transformation education has ever seen”.

In a 2023 TED Talk, he demonstrated how the Khan Academy’s AI assistant – Khanmigo – works as a virtual personal tutor, helping students practice problem solving, gain instant feedback on their work, and even simulate conversations with historical figures.

Other platforms, like ChatGPT’s Study Mode and Google’s LearnLM, are doing the same. They aim to help students learn how to find the answers, rather than just handing them out.

This matters because many young people are already using AI for schoolwork. And the early indications aren’t exactly pointing towards a utopian future.

A recent MIT study found that when students used ChatGPT to write essays, their brain activity plummeted. They “underperformed at neural, linguistic, and behavioural levels,” while those who worked unaided showed higher levels of creativity and took more satisfaction in their work. Psychiatrist Dr Zishan Khan warned that an over-reliance on AI “can have unintended psychological and cognitive consequences,” especially for developing minds.

Being able to ask ChatGPT anytime things get tricky may sound great, but in education, difficulty isn’t a bug – it’s a feature. Learning requires struggle. Education experts call this “desirable difficulty”.

As psychologist Dr Paul Penn puts it:

The lesson from the past few decades of research in cognitive psychology is that studying (and learning generally) is a bit like visiting the gym: if you want the best results, you have to sweat a bit.

The challenge for educators now – and for all of us – is to work out how we use the incredible power of AI to enhance the learning process, rather than replacing it.

How to teach yourself

Some of the greatest minds in history learned by themselves. Michael Faraday, son of a blacksmith, taught himself chemistry while working as a bookbinder’s apprentice. Joseph Priestley, who discovered oxygen, was largely self-taught. Benjamin Franklin left school at 10.

Today’s autodidacts have more resources and tools than ever: MIT OpenCourseWare, Yale’s open lectures, Coursera, Duolingo, Brilliant, Masterclass, not to mention all those entertaining YouTube explainers.

Cartoon version of Mike pointing at colourful framed posters with titles like "Magic Mushrooms" and "Placebo Paradox." Pink and white spiral background.

But motivation alone isn’t enough. It’s why nine out of ten people who started a MOOC didn’t finish. It might also be why that Harvard computer science lecture you bookmarked with the best of intentions is still gathering digital dust.

But there’s one old-school tactic that we know still works: structure. Substacker Parker Settecase even suggests designing your own “seasons of study”, just like a university course. Block out time, set goals and measure your own progress.

It’s an approach that complements what research from psychology says makes for successful and deeper learning.

Here are five recommendations from Dr Paul Penn on how to study effectively (taken from an extended essay on Psyche):

1. Space out your study sessions

Short, regular sessions beat last-minute planning.

2. Mix your subjects

Alternating similar topics helps you see distinctions.

3. Focus on understanding, not memorising

Question what you read as you’re reading it. It’s only when you can explain the subject in your own words that you truly understand it.

4. Test yourself

One recommended method is the 3R approach: read a passage of text, put it to one side, try to recall the information in your own words, and then review how accurate your recital was. Repeat the process until you’re satisfied you’ve got it.

5. Don’t just highlight… think!

Studies suggest that the more you highlight, the less benefit you get from it. Active engagement beats passive reading.

Yellow background with a graduation cap, pink downward graph, and text "ARE DEGREES LOSING VALUE?"

Will traditional qualifications still matter?

In The Abundant University, Michael D. Smith offers up a big reason why universities and colleges haven’t gone the way of Blockbuster:

We haven’t yet had a structural disruption in higher education because credentials remain scarce. As long as you need a degree to get a good job, the university is safe from disruption.

There are signs, however, that this may be changing.

Google, IBM, Amazon, and Microsoft are already offering their own certificates and training programmes – shifting from degree-based recruitment to skills-based hiring. Amazon has pledged $700 million to upskill 100,000 employees. Google’s Career Certificate can qualify candidates for jobs, with no degree required.

For many careers, a strong portfolio can speak louder than qualifications. Online platforms like HackerRank, Kaggle and TopCoder have become go-to places for companies to find talented coders and data scientists.

Some believe we may be heading for a future where skills speak louder than diplomas – where people can develop their talents, change careers or simply feed their curiosity without taking on the financial costs of higher education.

But it’s unlikely that traditional qualifications will ever become obsolete – just as theatre survived Netflix. We’ll still value what real classrooms offer that screens can’t: mentorship, collaboration, connection and community. But they’ll no longer be the only show in town.

The future – a new golden age of learning?

While many debate whether AI will destroy education, the better question is what we want education to be.

AI will almost certainly take on some of what teachers do today. But in doing so, it could free them to focus on what humans do best: nurturing curiosity, confidence, and connection.

As education innovator Anant Agarwal writes in The New York Times:

It’s possible to imagine a world where AI can ingest rich learner data and create personalized learning paths for students, all within a curriculum established by the teacher… Teachers can continue to be deeply involved in fostering student discussions, guiding group projects and engaging their students, while A.I. handles grading and uses the Socratic method to help students discover answers on their own.

For some pupils, this isn’t a speculative vision of the future – it’s already happening. At the Alpha School in Austin, Texas, pupils spend just two hours a day studying core subjects like maths and reading through AI-driven software that adapts to each child’s pace. The rest of the day is spent collaborating with classmates and teachers on practical skills such as public speaking, entrepreneurship, and financial literacy.

Like the printing press, the radio lecture, or even the Open University before it, AI represents the next great leap in how we learn – with the potential to make personalised, lifelong education available to everyone.