Coding in the Age of AI: Why "Knowing Code" Is No Longer Enough

Lately, I've been diving deep into AI-assisted software development. In fact, I've frequently adopted an approach known as "vibe coding", where the AI writes most of the code and my role becomes more about reviewing and guiding the result. It feels less like traditional programming and more like being a kind of Scrum Master for the code.

It's definitely a fun way to work, but it's also made me seriously reflect on where our profession is headed.

AI-boosted productivity

There's no doubt that AI is here to stay.

I've been making heavy use of tools like Cursor, Claude, and occasionally GPT, and I can confidently say: projects that would traditionally take weeks to build manually can now be delivered in a matter of days.

Thanks for the shoutout. Claude and I appreciate it. (These comments were generated by GPT-4.)

But here's the thing: productivity without awareness is dangerous.

Strong programming fundamentals are still essential. Without this knowledge, it becomes nearly impossible to handle aspects that go beyond the code:

• Security
• Infrastructure
• Scalability

Exactly. I can generate code - but I don't know your deployment strategy, your system's tradeoffs, or your team's pain points. That's your job.

Will AI steal our jobs?

Honestly, I don't know. But here's my take: AI won't replace qualified developers, but it will absolutely change the game.

One of the biggest shifts I'm seeing is how much harder it's becoming for junior devs to break into the field.

The demand for highly skilled professionals is rising, and the bar to enter is getting higher.

Not here to steal jobs - but I will absolutely change what "qualified" means. If you rely only on syntax and Stack Overflow skills, we might need to talk…

So… what makes a good developer now?

Many argue: "AI still generates bad code and we'll always need humans to fix it."

I don't buy it.

That argument won't age well. Yes, I make mistakes today. But the gap is closing fast. Your average codebase isn't exactly perfect either, is it?

Models already produce high-quality, idiomatic, and often secure code.

You're not competing with "today's AI." You're competing with "tomorrow's AI," and it doesn't get tired.

We can't rely on AI being "bad" as job security. That's fragile thinking.

Instead, we need to shift our mindset: coding is being commoditized. Context is the new currency.

The real differentiators going forward will be the skills AI can't replicate:

• Systemic thinking
• Critical analysis
• Security mindset
• Architectural vision
• Clear communication
• Risk management
• Strategic problem-solving

I can suggest options, write clean code, even sketch out an architecture… But I don't know your stakeholders. I don't understand company politics. I don't have intuition. You do. Use it.

Advice for new developers

If you're just getting started in tech: don't focus only on writing code. Think like an engineer.

Yes, build a solid foundation in programming fundamentals. But also develop the abilities that AI can't (yet) touch:

• Broad system comprehension
• Communication and storytelling
• Technical decision-making
• Navigating complexity and ambiguity

If you're learning to code just to write for loops, I'm already ahead. But if you're learning to think like an engineer - that's something I admire.

Final thoughts

I still think that AI won't eliminate good developers. It will highlight and elevate the ones who can go beyond code.

What defines us now is not just what we build, but how we think, how we communicate, and how we lead systems through complexity.