Forget the 10,000-hour rule. A Harvard study suggests AI can accelerate learning by 2.3x, which sounds like hype until you realise it’s not about AI being magical—it’s about having a tutor who never gets tired, never judges, and works on your schedule.
The catch: most people ask Claude or ChatGPT a question, get an answer, and move on. That’s search, not tutoring.
A tutor does something different. It breaks material into digestible pieces, tests you repeatedly, spots gaps, and forces you to explain things back in your own words. You can automate that. Here’s how.
The 20-hour sprint
Mastering something new doesn’t need 10,000 hours. It needs 20 hours of focused, intentional practice. The trick is picking the right 20 per cent of concepts—the ones that generate 80 per cent of results.
Ask Claude: “I need to learn [topic] in 20 hours. Create a plan focused on the 20% of concepts that generate 80% of the results. Divide it into 10 sessions of 2 hours each, including the best resources and a 15-minute review at the end of each session.”
You’ll get back a roadmap. Follow it, not Netflix.
Explain it back (the Feynman Technique)
Here’s where most people stop thinking they understand: if you can’t explain something, you don’t understand it yet.
Ask Claude to explain a topic. Then ask it to correct you when you explain it back in your own words. Point out your mistakes, re-teach what didn’t land, and repeat until you can explain it without help. The friction in this loop is where learning actually happens.
One page, five minutes
Before a meeting or test, you need everything on a single page. Bullet points, a diagram, an example. Not for memorisation—for recall under pressure.
Prompt: “Summarise the main concepts of [topic] on a single page. Use bullet points, diagrams, and examples so I can review it in 5 minutes.”
Print it. Read it twice. Done.
Test yourself until it sticks.
Believing you’ve learned something and knowing it are different things. A test shows the gap.
Prompt: “I’ve just studied [topic]. Ask me 10 increasingly difficult questions to assess what I actually understood. After each answer, give me a score and explain what I got wrong.”
Answer the questions. Read the feedback. Note the gaps. Move on to the next ten.
Build a learning ladder.
Expertise is a climb with five distinct levels. Level 1 is “I’ve heard of this.” Level 5 is “I could teach it.”
Prompt: “Break down [topic] into 5 difficulty levels. For each level, from beginner to advanced, show me what I should be able to do at that stage.”
This is your progress map. You’ll know exactly where you stand and what’s next. No mystery.
Find the five resources that matter
Not all resources are equal. Some are waste. Some are gold.
Prompt: “List the 5 best resources for learning [topic] quickly—books, videos, courses, experts. Explain why each one deserves my time.”
You’ll get back a ranked list with reasoning. Skip the rest. Ignore the noise.
The actual setup
One prompt, start to finish:
“I need to learn [topic] quickly. Here’s what I want: a 20-hour plan split into 10 sessions, the best resources ranked, a one-page study sheet, increasingly difficult test questions, a learning ladder showing my progress, and the Feynman Technique applied throughout. After each session, show me what I’ve mastered and what I haven’t yet.”
Claude will structure the entire thing. You execute.
The 2.3x improvement isn’t magic. It’s what happens when you stop reading passively and start using someone (or something) that won’t let you pretend you understand something you don’t. AI is reliable at that. Humans usually aren’t.
The tutor is ready. Whether you’ll actually use it properly is another question.
