So I was in an interview recently.
The interviewer, a pretty senior engineering director, took a sip of water and hit me with this:
“Do you use AI to write code? AI is already 100x faster than you and probably makes fewer mistakes. So what’s the real value of a programmer now? And where do you think this career is going?”
Yeah… not exactly a small question.
First: Let’s be honest
My answer started like this:
“AI is absolutely making coding cheaper. But that doesn’t mean building software is getting easier.”
Back in the day, our edge was:
- knowing syntax inside out
- mastering frameworks
- remembering all the weird edge cases
Now? AI wipes out most of that instantly.
So if “writing code” is becoming commoditized, we need to rethink where our value actually comes from.
Here’s how I see it 👇
What actually becomes MORE valuable
1. Understanding real-world business context
AI can turn a vague idea into a 30-page spec.
But it doesn’t understand:
- what your boss really wants
- team politics
- business trade-offs
Turning messy, conflicting requirements into something buildable?
That’s still very human.
2. Making real-world trade-offs
AI loves perfect architectures.
But it doesn’t know:
- your budget is $5000
- your team can’t handle Kubernetes
- your legacy database is untouchable
The “right” solution is always contextual.
And context is where humans win.
3. Taking responsibility (this one is huge)
AI can review code.
AI can generate code.
But when production crashes, data leaks, or money is lost…
AI doesn’t get fired.
AI doesn’t go to court.
Someone has to own the outcome. That’s us.
So my conclusion was:
Let AI handle the execution. We focus on decisions, context, and accountability.
If that’s true… what should we double down on?
Here are 5 things I think will matter a lot in the next few years:
1. From coder → system thinker
Stop thinking in functions and APIs only.
Start thinking:
- system design
- trade-offs
- scalability vs cost
AI is great at local execution, terrible at global decisions.
2. Treat AI like a power tool, not a crutch
Bad usage:
“Generate code → copy → doesn’t work → blame AI”
Better usage:
- give rich context
- ask for design first
- iterate with constraints
Your prompting skill = your leverage.
3. Code review becomes a superpower
Honestly:
People who can review AI-generated code well will be extremely valuable.
AI code often looks clean… but hides:
- logic flaws
- edge case failures
- long-term maintainability issues
You need sharp judgment, not just speed.
4. Think more like a product person
Pure “task executors” will struggle.
Engineers who understand:
- users
- business goals
- trade-offs
…will stand out a lot more.
5. Actually apply AI to real problems
Not building models.
But using AI to:
- fix internal tools
- improve workflows
- automate painful processes
That’s where real value is created.
Different stages, different strategy
Quick thoughts:
- 0–3 years: Don’t let AI kill your fundamentals
- 3–7 years: This is the danger zone. You must level up to system design
- 7+ years: Move into strategy, architecture, and org-level impact
- Domain experts: This might be your golden era
Final thought
The most valuable engineers won’t be the fastest coders.
They’ll be the ones who:
- understand problems deeply
- design resilient systems
- use AI as leverage
- and take responsibility for outcomes
I’m curious:
How are you adapting to AI in your daily work?
Do you feel it’s replacing parts of your job, or amplifying you?
Would love to hear different perspectives 👇
Top comments (0)