Learning with AI

AI changed how we learn. This session covers how to use it to accelerate your growth as an engineer.

Pre-Session Video

Watch this before the session — it covers the core thinking behind everything we’ll do.

The Shift in Learning

The best engineers aren’t the ones who know the most — they’re the ones who learn the fastest. AI tools have changed the game: you can now learn in days what used to take months.

But there’s a right way and a wrong way to use these tools. The difference determines whether you actually grow as an engineer or just produce code you don’t understand.

1. Learn by Building, Not Just Watching

Tutorial hell is real. Watching someone code is not the same as coding. You can watch 100 hours of React tutorials and still not know how to build a real app from scratch.

Why this happens: Watching gives you the illusion of understanding. You follow along, everything makes sense in the moment, but when you sit down alone, you freeze. That’s because understanding code is not the same as being able to write code.

The fix: Use AI to scaffold a project, then break it. Change things. See what happens. Delete a function and try to rewrite it. Rename a variable and watch the errors appear. That’s how you actually learn.

The Break-It Method

1. Ask AI to build something simple — a form, a landing page, an API endpoint
2. Read through the code and ask it to explain every line
3. Delete half the code
4. Try to rebuild it from memory, referencing the original only when stuck
5. Repeat until you can build it without help

This forces your brain to encode the knowledge, not just recognize it. Recognition is easy, recall is hard — but recall is what you need when building real projects.

2. Use AI as a Tutor, Not a Crutch

There’s a difference between using AI to understand code and using it to avoid understanding code.

The crutch approach looks like this:

“Write me a React component”

The tutor approach looks like this:

“Explain how this React component works, line by line, and why you chose useEffect over useState for this specific case”

The first gives you output. The second gives you knowledge.

How to Prompt for Learning

The Why-After-How Pattern

Every time AI gives you an answer, ask “why” at least once:

• “Why did you choose this approach over alternatives?”
• “Why does this line work when the other one didn’t?”
• “Why is this considered best practice?”

This is how you build mental models — frameworks for thinking about problems that transfer to new situations.

3. Build a Feedback Loop

The fastest way to improve is to get feedback quickly.

The loop: Code → Build → Run → See → Fix → Repeat

When you can complete this loop in minutes instead of hours, you learn 10x faster. Here’s why:

• Immediate feedback reinforces correct mental models before they solidify wrong ones
• Small iterations make it easier to identify what changed and why
• Rapid experimentation lets you test hypotheses about how code works

How to Speed Up Your Loop

1. Work in small files. Don’t build a 500-line file before testing. Build 20 lines, test, repeat.
2. Use hot reload. Most frameworks have it. If yours doesn’t, set it up — it’s worth the time.
3. Console.log everything. Don’t guess what’s happening. Log it. See it. Understand it.
4. Deploy early. Don’t wait until your project is “done” to put it online. Deploy on day one, deploy every day after.

4. Keep a Learning Journal

After each session, write down:

• What you learned (one specific thing, not “React is cool”)
• What confused you (be honest — confusion is where growth happens)
• What you want to learn next (follow your curiosity)

Even a single sentence per day compounds into deep knowledge over time.

Why This Works

Your brain is bad at remembering what you learned last week. A journal is a search engine for your own knowledge. Three months from now, you’ll be able to look back and see exactly how far you’ve come.

It also forces you to articulate what you know. If you can’t explain it simply, you don’t understand it well enough.

5. The 80/20 of Learning to Code

80% of your results come from 20% of your efforts. Here’s where to focus:

High-Impact Activities

• Building real projects — not tutorials, not exercises, real things people use
• Reading other people’s code — open source, GitHub repos, your classmates’ projects
• Debugging broken code — nothing teaches you more than fixing something that doesn’t work
• Explaining concepts to others — teaching is the ultimate test of understanding

Low-Impact Activities

• Watching tutorials without coding along
• Memorizing syntax instead of understanding patterns
• Copying code without understanding it
• Switching frameworks every week

The Minimum Viable Learning Stack

You don’t need to learn everything. For web development, start with:

1. HTML/CSS — structure and styling
2. JavaScript — the language of the web
3. One framework — React, Vue, or Svelte (pick one, stick with it)
4. Git — version control (you’ll need this for everything)
5. One deployment platform — Cloudflare Pages, Vercel, or Netlify

Master these five things and you can build almost anything. Everything else is optimization.

Key Takeaway

The best engineers are professional learners. They’ve mastered the skill of acquiring new skills. AI didn’t change what you need to learn — it changed how fast you can learn it. Use it wisely, and you’ll be building real projects in weeks, not months.