You're Not Replaced Yet — That's Exactly Why You Should Be Worried

Still employed? Good. Still getting paid the same? Even better. Still doing largely the same job you were doing two years ago? That last one is the problem.

The reason most people feel safe from AI right now isn't because they actually are. It's because the damage doesn't announce itself. Nobody sends a calendar invite titled "Your role is becoming obsolete." The shift is quieter than that — and it's already well underway.

The slow disappearing acts

When people imagine AI taking jobs, they picture a dramatic moment. A pink slip. A press release. A robot literally sitting down at a desk. What's actually happening is far less visible and far more effective.

Companies aren't firing their existing staff en masse. They're just not backfilling. When someone quits, the role sits open — or gets quietly absorbed by a smaller team running AI tools. When a department needs to expand, the headcount request gets denied because the output is already there without the hire. Positions that would have existed two years ago simply don't get created.

The job market feels the squeeze before any individual worker does. And by the time it's personal, the window to adapt has already been open for a while.

"You don't get fired on a Tuesday. You just notice, slowly, that fewer people are doing more work — and none of them are new."

Why "still employed" is not the same as "safe"

There's a version of job security that's real: your company needs you, your skills are genuinely hard to replace, and the work you do requires judgment that no tool has yet. Then there's a version that's borrowed time: you're still there because the transition hasn't hit your department yet, your manager hasn't made the call, or the tool isn't quite good enough — but it will be by next year.

The gap between those two versions is closing. And most people haven't asked themselves honestly which one they're in.

According to a Harvard Business Review analysis, many US companies are already cutting roles not because AI has fully proven itself — but because leadership believes it will. The anticipation alone is reshaping org charts right now. You don't have to be replaced to feel the effects of an industry that's restructuring around the expectation that you will be.

The roles that feel safe but aren't

Mid-level managers who run meetings, track deliverables, and report upward. Junior analysts who clean data and build slide decks. Marketing coordinators who write briefs, schedule posts, and pull performance numbers. Account managers who send follow-up emails and maintain CRM records. These roles feel embedded in how companies run. They also happen to describe what AI handles well.

That doesn't mean every person in those roles is gone tomorrow. It means the next version of those roles — the one that gets hired for in 2027 — looks different. It requires less of the work AI does and more of the work it can't: genuine client relationships, hard judgment calls, creative thinking that comes from actual experience.

As CNBC reported, economists warn there is "much more in the tank" when it comes to AI pressure on white-collar work — with major banks, retailers, and tech companies already signaling that the size and shape of their workforce will look fundamentally different within a few years.

The gap between now and too late

This is the part that doesn't get enough attention. The workers who come out ahead of this shift won't be the ones who panic in 2028 when a layoff round finally reaches them. They'll be the ones who spent 2025 and 2026 being honest about which parts of their job are on borrowed time — and deliberately moving toward the work that isn't.

That means identifying what you do that requires real human context. Not just skills that sound human, like "communication" or "relationship management," but actual work where the outcome changes based on who specifically is doing it. That's the thread worth pulling.

It also means getting comfortable with AI tools now — not to be trendy, but because workers who understand how to use them are a layer above the tools themselves. They're harder to cut and more valuable to keep.

The one question worth sitting with

If your company hired someone new into your role tomorrow and gave them access to the same AI tools everyone else has, how long before they could do most of what you do?

If the answer makes you uncomfortable, that's the right response. Not because your job is gone — but because that discomfort is the only thing that leads to actually doing something about it while there's still time.

Last Publication: Prompt Engineering Had a Good Run. Now What?