13 · Learn Anything Faster with AI

AI can be the best personal tutor you have ever had, or a machine for feeling productive while learning nothing. The difference is how you use it. This page combines proven learning science with AI techniques so you grow faster in anything: languages, frameworks, domains and soft skills.

The trap: outsourcing your thinking

If AI writes every line and you only accept it, you get output without understanding. It feels fast, but your skills stop growing, and you become unable to judge whether the AI is right. That makes you worse at using AI.

The learning loop

Six learning techniques, amplified by AI

1. Active recall

Retrieving information from memory strengthens it far more than re-reading.

I just studied Scala 3 givens and using clauses. Quiz me with 5 questions, one at a time, from easy
to hard. Wait for my answer each time, tell me if I'm right, and explain what I missed.

2. Spaced repetition

Review at growing intervals (1 day, 3 days, 1 week, 1 month).

Turn these notes into 15 flashcards (question / answer), one concept per card, in a format I can
import into Anki (front;back per line). Prefer "why" and "when" questions over definitions.

3. The Feynman technique

Explain the concept simply. Where you stumble is where your gaps are.

I'll explain how attention works in Transformers as if to a beginner. Act as a curious student:
point out every vague or wrong part of my explanation and ask follow-up questions.
My explanation: ...

4. Elaboration and analogies

Connect new knowledge to what you already know.

Explain Python's asyncio event loop by comparing it with Cats Effect fibers, which I know well.
Where does the analogy break down?

5. Deliberate practice

Practise just beyond your current ability, with immediate feedback.

Give me a Scala exercise on tail recursion and foldLeft that is slightly harder than FizzBuzz.
Don't show the solution. When I submit, review it like a senior engineer and give me a harder variant.

6. Interleaving

Mix related topics instead of drilling one in a block. It improves your ability to pick the right tool.

Give me 6 short problems mixing Option, Either, Try and exceptions, in random order. For each
I must choose the right type and justify it before writing code.

AI tutor roles

Playbooks

Learn a new programming language or framework in two weeks

1. Day 1 – Map: "I'm an experienced Scala developer. Give me a map of Rust: the 20% of concepts that cover 80% of daily use, what will feel familiar, and what will surprise me."
2. Days 2–5 – Chunks: one concept per day: explanation → tiny runnable example → exercise → quiz.
3. Days 6–10 – Build: a small real project (CLI tool, REST service). Write it yourself, with AI as reviewer and hint-giver.
4. Days 11–13 – Read real code: pick a well-known open-source project and have AI explain modules as you read them.
5. Day 14 – Teach: write a one-page "X for Scala developers" note. Ask AI to find errors in it.

Onboard onto an unfamiliar codebase

1. Ask the agent for an architecture overview: modules, data flow, entry points. Ask it to draw an ASCII or Mermaid diagram.
2. Trace one real request end-to-end: "Follow a POST /orders from route to database. List each file and function."
3. Ask "what would surprise a newcomer here?" and "where is the riskiest code?"
4. Make a small change yourself. Use AI to review it against the project's conventions.
5. Write what you learned into AGENTS.md / docs. That helps the next human and the next agent.

Learn a new business domain

I'm a software engineer joining a team that builds payment-reconciliation software.
1) Explain the domain in plain terms: actors, money flows, key documents.
2) Give me a glossary of the 25 terms I'll hear in the first week.
3) What are the classic failure cases and edge cases engineers in this domain must handle?
4) Suggest 5 questions I should ask domain experts in my first week.

Read a technical paper or long document

Help me read this paper in three passes.
Pass 1: the problem, the main claim, why it matters (5 bullets).
Pass 2: the method explained with an analogy and a small example.
Pass 3: limitations, assumptions, and what I'd need to check to trust the results.
Then quiz me on the key ideas.

Build a personal growth system

• Learning log: a daily markdown file with three lines: what I learned, what confused me, what I'll try tomorrow. Once a week, ask AI to spot patterns and suggest focus areas.
• Personal knowledge base: keep notes in your own words, in Markdown, in git. You can later make them searchable with RAG (Level 2).
• Prompt library: save prompts that worked. Turn the best into skills (page 07).
• Monthly skill review: "Here is my learning log for the month. Assess my progress against the Level 1 objectives and propose next month's plan."
• Teach to learn: explain topics to colleagues or friends, or write short posts. Teaching is the strongest test of understanding.

## 2026-10-08
- Learned: Scala 3 opaque types have zero runtime cost; companion holds smart constructors.
- Confused: when to use `inline given` vs `given`.
- Tomorrow: quiz myself on givens; read the Scala 3 reference page on contextual abstractions.
- AI used for: explanation + quiz (learning mode), test scaffolding (delivery mode).

Personal prompt pack for learning

"Explain <topic> at three levels: intuition, practical, deep. Stop after each and ask if I'm ready."
"What are the top 5 misconceptions people have about <topic>?"
"Give me a minimal runnable example in Scala 3 and the same in Python."
"Here's my understanding: <...>. Rate it 1-10 and correct it."
"Design a 30-minute hands-on exercise for <topic> with a hidden twist."
"What should I learn next after <topic>, and why? Give me a dependency graph of concepts."
"Summarise today's session into 5 flashcards and 3 open questions."