A Quick Recovery Guide for AI-Dependent Coders

AI Summary5 min read

TL;DR

AI에 지나치게 의존하는 코더들을 위한 회복 가이드: AI 사용을 금지하고, 언어 구문과 표준 라이브러리를 공부하며, 직접 프로젝트를 구축하고, 오류를 스스로 해결하는 방법을 제안합니다.

Key Takeaways

•AI 사용을 일시적으로 금지하고, 계산기처럼 기초를 익힌 후에만 활용하세요.
•주 언어의 구문과 표준 라이브러리를 공부하며, 예제를 직접 타이핑해 실습하세요.
•SQL을 학습하고, 작은 프로젝트를 처음부터 구축하여 실전 경험을 쌓으세요.
•오류 발생 시 AI 대신 구글 검색이나 고무 오리 디버깅으로 스스로 해결하는 습관을 기르세요.
•공식 문서를 읽고, 데이터 구조와 컴퓨터 과학 수학을 공부하여 문제 해결 능력을 강화하세요.

0. Ban AI.

Think of AI as calculators in math classes. You can't use them until you know the procedure you want to automate by hand.

Like any mom disciplining her kid, "No more AI until you do your homework..."

1. Study your main language syntax.

Get to know the syntax of your language of choice: write variables, functions, loops, classes...

For that, grab a textbook or watch a "all you need to know about X in 4 hours" YouTube videos. But don't just passively consume them, recreate the examples and projects from them. By typing them out. No Control C + Control V.

2. Know your standard library.

Get familiar with your standard library:

Write a variable and see what your editor or IDE suggests.
What methods can you use with that type?
Look at their signature and docstring.

3. Study SQL.

No matter how powerful ORMs are, we can't escape from SQL.

We've had SQL since the early 70s and chances are we're still using SQL another decade or two.

Learn to create tables, write queries aggregating results, and learn about JOINs. Download a light version of the StackOverflow database and play with it, if you want realistic examples.

4. Build a toy project from scratch.

OK, I'm not talking about reinventing the wheel to write your own text editor or something.

I'm talking about building a recipe catalog, a todo app, or a CLI wrapper around a free API. And build it from scratch: right click, then create new folder in your editor or IDE, and so on. It will teach you a lot.

5. Find your own answers.

When you get an error message (you will if you follow #4), resist the urge of going back to AI or simply asking a friend.

Try to figure out errors and exceptions on your own. Start by googling the error message. There's a thing called Google that finds web pages with answers to our questions. Sure, it's old-school but it builds real muscle. Remember, AI is still banned. (See #0)

6. Learn the most common data structures.

80% of the time, you'll only need lists and dictionaries. But there are more data structures of course.

Learn to use them and how to implement them. You won't have to implement them from scratch at your daily job, but it will stretch your problem-solving muscles.

7. Study a textbook on Math for Computer Science.

Unless you're working on niche domains, you won't need advanced Math.

But grab a book on Discrete Math (or Math for Computer Science) and study a chapter or two. Again, to sharpen your thinking.

8. Practice rubber-duck debugging.

You will get stuck a lot. That's a feature of being a coder, not a bug.

When that happens,

Grab pen and paper
Go through your program line by line
Talk out loud

9. Read the official documentation.

Pull out the Mozilla's Web Docs, Microsoft Learn, and any other official source for your language of choice, and not only read it, but come up with your own examples and think of how you can use what you're reading in your own code.

AI is a blessing for learning. Ask any veteran who learned from reference manuals, language specifications, and magazines, they'll tell you. Just don't let AI think for you.

OK, let's slightly lift the AI ban, don't use it to generate code. Use it as your copilot, not as your captain.

To help you build hype-proof skills, I wrote Street-Smart Coding. The roadmap I wish I had on my journey from junior/mid-level to senior.