AI Is Not Going to Take Your Job. But This Person Will.

A few weeks ago I wrote about entry-level hiring and what AI is actually doing to the pipeline. A friend pushed back — fairly — and said the real concern isn't the junior folks. It's the seasoned professionals. The people who've spent years, maybe decades, building real expertise. Are they at risk?

He has a point. But it points to the same underlying question.

The issue isn't really about where you sit on the org chart. It's about posture. Whether you're treating AI as something that's happening to you, or something you're deliberately choosing to leverage.

Two Ways to Think About AI at Work

There's a meaningful difference between two approaches, and most people haven't consciously chosen which one they're taking.

The first is what I'd call +AI. You take your existing task list — the things you've always done, the way you've always done them — and you look for places AI can speed things up. Drafts get written quicker. Research gets pulled faster. Reports get formatted more efficiently. Optimization. Real and valuable.

The second approach starts with a different question: How does what I do add value? Not what do I do — but what value does it create for my employer, my client, my customer? And then: is there a way to leverage AI to deliver that value so I can spend my time adding value on top of that? I call this AI+.

There's a difference between asking how AI can help you do what you already do faster, and asking how AI changes what you do — the value you create, the problems you can solve. One makes you more efficient. The other makes you harder to replace.

NVIDIA's 2026 State of AI report surveyed over 3,200 companies and found that 88% say AI has already increased annual revenue, with 30% seeing greater than 10% growth. More telling: 53% said the single biggest business impact was improved employee productivity. The data reflects what the smartest operators are already acting on.

The Irreplaceable Gap

Based on OpenAI's GDPVal ..... Generative AI gets things right somewhere in the range of 83 percent of the time depending on the domain, that's up from 70% only a few cycles ago. Impressive. However if you flip that, it's a 17 percent error rate, on a good day. And those errors are usually confident-sounding, well-formatted, and plausible. They read like they're right.

This is where the seasoned professional earns their seat.

Knowing what questions to ask. Knowing how to prompt for what you actually need. Knowing when something doesn't add up, not because you ran it through a fact-checker, but because years of experience made the wrongness visible to you immediately. That's not a skill AI has. That's a skill you built over years of doing the work.

The human-in-the-loop isn't there to rubber-stamp what the model produces. They're there to catch what it gets wrong, interpret what it gets right, and communicate the output in a way that creates real value. That loop — prompt, evaluate, refine, apply — is where your expertise lives now.

The Tool That Teaches Itself

Jensen Huang made a point recently that I keep coming back to. He said AI is unique among disruptive technologies in one specific way: if you don't know how to use it, you just ask it. You ask the tool how to use the tool, and it tells you.

Compare that to every other tool in history. A Cuisinart is genuinely useful, but you have to figure out the interface — read the manual, learn the settings, figure out how much to put in and how to adjust the cuts. There's a ramp. You climb it alone.

AI doesn't work that way. It meets you where you are. If you're new to it, it walks you through. If you're experienced, it goes deeper. It's the first tool in history that removes its own learning curve.

I spent the better part of my career as a developer. Over time, as I moved into leadership, the writing-code muscle atrophied. The syntax stopped being automatic. The mental pattern recognition that used to be effortless started taking longer. But AI changed that. I use these tools to jumpstart what I'm building — and the moment I see the output, I recognize it. The knowledge was always there. The tool surfaced it. It made me a builder again.

That's not a story about AI replacing expertise. That's a story about AI restoring it.

The Person Who Will Take Your Job

Here's the honest version of the AI-and-jobs conversation.

AI is not going to take your job. But someone who has your skill set and a real mastery of AI will take your job. That's the actual threat — not the model, but the person who figured out how to leverage the model while still bringing the human judgment, domain experience, and contextual wisdom no model can replicate.

Whoop CEO Will Ahmed understood this when he announced the company was hiring 600 people — nearly doubling its 800-person workforce — while investing heavily in AI. His quote: "Right now, companies are debating whether to hire more people or just invest in AI. We are doing both." The companies getting this right aren't asking how many people they can cut. They're asking what they can build now that they couldn't before.

Optimization vs. Evolution

At this point, not using AI in some meaningful capacity is roughly equivalent to saying you don't use a computer. It's not a principled stance, it's a liability. But clearing that bar isn't the goal either.

Optimization asks: how does AI help me do what I already do, faster?

Evolution asks: how does AI change what I do — the value I create, the problems I can solve, the contribution I can make?

Those are different questions. The person asking the first one is getting more efficient. The person asking the second one is building something you can't easily catch up to.

Which question are you asking?