I keep seeing this question pop up: is AI stealing our jobs?
Here’s what I’ve learned from watching this unfold: the divide isn’t between humans and AI. It’s between people who figured out how to use AI as a force multiplier and people who got replaced by someone who did.
The multiplier effect
Most people are looking at this wrong. AI doesn’t create a zero-sum game. It creates an exponential one.
I’ve seen engineers who learned to use AI effectively do what previously took a team of five in a fraction of the time. That’s not about replacing jobs. That’s about changing what’s possible.
The people succeeding with AI aren’t the ones who stopped thinking. They’re the ones who started thinking differently. They frame problems clearly before asking AI for solutions. They understand the constraints that AI might miss. They verify the output with their own expertise.
Garbage in, garbage out
This isn’t just a technical principle anymore—it’s a career principle.
If you bring AI vague requirements, unclear thinking, and no domain expertise, you’ll get garbage output. Fast. The AI amplifies your confusion, not eliminates it.
But if you bring clear thinking, domain knowledge, and specific direction? That’s when you see the multiplier effect. The AI scales your expertise, not replaces it.
Vibe coding vs. agentic engineering
Karpathy recently drew a distinction that’s becoming critical. There’s “vibe coding” and there’s “agentic engineering.”
Vibe coding is when you have a rough idea, ask an AI to build it, and hope for the best. You get something that looks functional until you try to use it in production.
Agentic engineering is when you understand system architecture, know what the AI needs to succeed, can spot where it’s going wrong, and can steer it toward the right solution. The AI becomes your most powerful tool, not your replacement.
The engineer’s advantage
This is where trained engineers have a massive advantage. They don’t just know how to use AI—they know what to ask for, where the pitfalls are, and how to verify the output.
A developer who understands system architecture can look at AI-generated code and immediately spot the architectural debt. Someone who’s debugged production systems knows what questions to ask the AI before it commits to the wrong approach.
The AI can generate code 10x faster, but only if you know what good code looks like and where the traps are.
The co-pilot mindset
The AI isn’t your replacement. It’s your most demanding co-pilot—one that needs clear instructions, domain knowledge, and someone who can spot when it’s about to fly into a mountain.
You have to treat it as a co-pilot, not as a slave to go do think execute your vision that you haven’t even shared with it. The AI can scale and accelerate your work, but you have to have the vision first.
What this means for your career
The jobs that will disappear aren’t the ones that AI can do. They’re the ones where humans do the work without understanding what they’re doing.
The jobs that will thrive are the ones where you combine human judgment with AI’s execution power. The ones where your expertise tells the AI what to focus on, where to be careful, and how to verify the results.
That’s not job stealing. That’s job evolution.
The question isn’t whether AI will change your career. It’s whether you’ll be the one steering the change or being changed by it.