Why Good Employees Lose First: The Uncomfortable Math Behind AI Layoffs

Nathan Chacón spent five years building a freelance translation practice the careful way: steady clients, near-flawless turnaround, a reputation good enough that work found him instead of the other way around. Beginning in 2023, he noticed something new — more of his regular clients quietly translating documents themselves with AI instead of sending them his way. He hadn’t gotten worse at his job. The market had simply stopped needing someone to do it the old way, one document at a time.

A translator interviewed by journalist Brian Merchant in 2025 put a sharper point on it. Fourteen years into an Italian-to-English freelance career, this translator had just had the best year of their working life in 2024 — fully booked, turning away clients, income at an all-time high. Then June 2025 arrived. Not one work request came in. Not one. The falloff wasn’t gradual. It was a wall.

Neither of these translators got lazy. If anything, they were better at their jobs in 2024 than they’d ever been. That’s what makes their story worth sitting with, because it breaks the assumption most of us grew up on: that being good at your job is what protects it.

The Uncomfortable Mechanism

In July 2025, Microsoft researchers published a study ranking occupations by how applicable current AI systems were to the actual work involved. Translators and interpreters landed at the very top of the entire list — more exposed than accountants, more exposed than paralegals. Not because translation is easy. Because it’s specifiable. A finished translation has a clear input, a clear output, and a standard by which quality can be judged without needing to have been in the room when it was produced.

A 2024 UK survey of professional writers and translators found more than a third had already lost work to generative AI, and 43% reported their income had dropped — not because clients found better translators, but because clients stopped needing a human step in the process at all.

The Replaceable Process Trap

Here’s the trap in its plainest form: you can be excellent at your job while your job is quietly built on a process that shouldn’t exist anymore, and your excellence will not save you. The thing your company was paying for was never really your care or your years of practice. It was the process running to completion, reliably, at a price the company was willing to pay.

Bank tellers who’d mastered every deposit-slip edge case weren’t protected when ATMs arrived, because their mastery had never been the point — throughput was. The pattern repeats with AI, one layer up the skill ladder. This is the part that stings for conscientious people: the more completely and reliably you execute a process, the easier you make it to document, copy, and hand to a machine. Diligence produces exactly the kind of clean, well-labeled output that makes a task look “solved.”

Being diligent inside a process that shouldn’t exist anymore isn’t protection. It’s the fastest way to document exactly how to replace you.

Are You Accidentally Training Your Replacement?

A quick, honest mirror — not a test to study for:

• Could you write out the exact steps of your last completed task as a checklist someone else could follow without asking a single question?
• Has a manager ever asked you to document “how you do X” so the process lives somewhere other than in your head?
• Is your output judged mainly by how closely it matches a template or established standard, rather than by a judgment call only you could have made?
• Has your company purchased, piloted, or mentioned an AI tool aimed at the exact category of task that fills most of your week?
• If you took a two-week vacation with zero coverage, would the backlog simply wait, or would something break that genuinely required your judgment?

A pattern of “yes” answers doesn’t mean you did anything wrong. It means you’ve been excellent at your job in exactly the way that makes a job easiest to automate. Noticing that difference — between value that comes from running a process well, and value that comes from being the person a process can’t fully contain — is where the real work starts.

Recommended Reading

This self-diagnostic is adapted from a full book built around exactly this distinction:

• AI Will Steal Your Job – Here’s How to Steal It Back, by Palatino D. James. Includes the full fifty-question Career Survival Assessment this diagnostic previews.

Browse the full Strong Through Change library →