How I use AIs to Learn Faster Than Most

FUTURE
How I use AIs to Learn Faster Than Most

Using AIs to learn is probably the most important Generative AI use case, and it’s not even close. In my case, it has completely transformed my learning process, and this is coming from a guy who literally earns his living from learning hard things fast.

I’m probably the fastest learner you’ve ever met, not because I’m a genius, but because I lean on AI more than anyone you know. Unironically, my competitive moat is nothing but better AI use than others.

Today, I’m going to give you my learning tricks while also giving you a sense of the profound transformation AI will have on modern education. This is no small joke; it’s a trillion-dollar opportunity that people like Sam Altman are counting on to make OpenAI a trillion-dollar company.

From Bloom’s two-sigma method to the legendary Skunk Works, we are laying the foundation for reforming the education system and your approach to learning.

You’re learning a lot today (pun intended).

The Thesis for Education Reform

My ‘AI education reform’ is based on two big ideas:

1. Aristocratic tutoring

2. Knowledge commoditization

Let’s unpack them.

Aristocratic tutoring

One of the most legendary figures in history is Alexander the Great, a young, brave Macedonian prince who decisively defeated the Persian king Darius III in the battle of Gaugamela in 331 BCE. This battle was the turning point in the war, eventually conquering the entire Persian Empire.

But probably the most remarkable point of his legend was his age, as he was just 26 when that battle took place, and just 22 when he crossed the Hellespont to start the War.

But he’s hardly the only example of extreme prodigious behavior:

• Mozart, the musical prodigy, was touring Europe and composing unique pieces (more than 800 in his entire life) at just five years old.

• John Stuart Mill was able to read Greek texts at just three

• Blaise Pascal published his “Essay on Conics” at sixteen,

• Ada Lovelace, credited with writing the first-ever ‘computer program’, did so in her late twenties.

And what all of them had in common? You guessed it, 1-on-1 tutoring.

The research behind the effectiveness of this method is undeniable, supported by both historical figures and recent research and implementations.

In the famous 1984 study by Benjamin Bloom, it was observed that average students who receive one-on-one tutoring in subjects to “mastery level” (i.e., requiring a minimum of 90% rather than 72% to advance) score roughly two standard deviations higher than students in conventional classrooms, placing them at the 95.44% percentile (better than 95.44% of students):

This study was particularly relevant because it was carried out on “average“ students, meaning that attributing genius capabilities to the aforementioned historical figures does not explain the entire picture, and aristocratic tutoring really played a vital role.

Thus, the value of this teaching method has been known for decades (or centuries), yet the issue with one-to-one tutoring is that, as the name suggests, it simply wasn’t feasible to deploy to millions of kids; it’s simply too expensive.

Luckily, AI changes that, particularly Large Language Models (LLMs). While there are only so many human teachers, the potential existence of AI tutors means anyone with access to an AI model (billions of people already) could benefit from personalized AI tutors.

But you don’t have to take my word that AI has potential here; we have proof. This proof is Alpha School, an Austin, Texas-based school whose kids are scoring in the top 0.1% nationally using AI tutors.

I already talked about this particular case previously, so I’ll cut to the chase. Alpha School has a unique approach to schooling:

• Only two hours of hard-school teaching, with each kid having their AI tutor and a “mastery goal” to achieve, all under the guidance of a human “guide” who leads the class.

• The rest of the day is spent on social exercises, ranging from building a drone to competing in a lemonade stand challenge.

• Two AI software tools, Incept (the AI tutor, built on frontier LLMs) and Timeback, another set of AI vision models that monitor and record the kid’s movements, like rushing through problems or spinning in the chair, going into a “waste meter” that highlights how much the kid they are wasting, represent the ‘AI component’ of all this.

• It’s incentives-based. The sooner the mastery level is achieved, the faster they can start playing. This is a very clear example of the heavy incentive-based mechanism applied by Alpha School. As Brian Holtz, the school’s co-founder, explains: “kids must love school so much they don’t want to go on vacation,“ and treating learning as an incentives-based system seems to be working incredibly well.

But this seems to be just another way for “aristocrats” to have an even bigger advantage over poor kids. This screams elitist. Yet, AI tutors have also proven to be a great way to reduce learning inequality.

In a 2024 World Bank-backed pilot in Edo State, Nigeria, students used GPT-4 as a virtual after-school tutor. The outcomes were astounding: after just six weeks, students showed learning gains equivalent to nearly two years of typical schooling.

They significantly outperformed a control group in English (the focus subject) and in general knowledge and digital skills. What’s more, these students even did better in their regular end-of-year exams across other subjects not taught by the AI.

Notably, girls benefited the most, not only helping to close the gender gap but also with female students improving even faster and catching up to the boys.

In a nutshell, tutoring seems to be a much better learning method, and AI unlocks this learning superpower by making it potentially accessible to all kids.

And what’s the other main idea?

Knowledge Commoditization

This is probably the hardest pill to swallow for many. With AI, knowledge is just for the sake of it; knowing a lot doesn’t automatically translate to economic value.

Don’t get me wrong, possessing knowledge is as important as always, because knowledge and experience drive better decision-making and are necessary tools to build stuff, but not as a valuable economic skill in itself, but as a means to an end.

And that ‘end’ is to act on that knowledge. I’ve said it many times: in an AI-rich world, it’s not about what you know, but what you do with what you know.

AI’s disruption is in itself ironic, because it speeds up learning tremendously while also making it less valuable overall.

Considering services like consulting or legal services, their value on a timely basis will decrease tremendously over the next few years as AI progressively reduces “time to insight”. We are literally living in a world where a CIO or COO might get enough insights from a 20-minute deep-research query to ChatGPT than an entire week’s pay to McKinsey (honesty, that speaks worse about McKinsey than it does well about ChatGPT).

But once reality hits, it’s going to be a massacre. This example alone is proof that you can drop a five-figure contract to basically a few thousand bucks, and not because I’m a genius, that workflow is quite frankly entirely out-of-the-box AI.

This four-hour workflow was probably billed at $100k in the US (Spain has a considerable pricing discount), and I mostly built it in a day (I would need maybe a week to make it production-ready, but not eight weeks as some consulting companies in that proposal claimed).

But I digress. The issue is that our modern education system is not adequately prepared for this paradigm shift; we continue to reward students for accumulating knowledge, but little for applying it effectively, assuming that knowing is sufficient to secure a living.

Put simply, we are still training our kids to be management consultants, lawyers, and investment bankers when we should be teaching them to be builders.

To summarise, AI introduces two democratization effects on education: one that democratizes aristocratic tutoring to the masses, and the other that democratizes the value of knowledge, making much of our financial services sector extremely overvalued today (but not for long).

So, how can we reform the education system? And, crucially, how do I get the most out of these models to put me ahead of everyone else?

Become a paying subscriber of Full Premium package to get access to this post and other subscriber-only content.

Already a paying subscriber? Sign In.

A subscription gets you:

• • NO ADS
• • An additional insights email on Tuesdays
• • Gain access to TheWhiteBox's knowledge base to access four times more content than the free version on markets, cutting-edge research, company deep dives, AI engineering tips, & more