My sister Sarah is one of the smartest people I know.
She spoke multiple languages fluently. She’d built a career as a freelance translator over 15 years — the kind of work that felt safe, skilled, deeply human. The kind of work that requires cultural nuance, tone, lived experience. Things machines weren’t supposed to understand.
Then, almost overnight, the work stopped coming.
No dramatic firing. No meeting with HR. No final paycheck. Just — silence. Clients who had relied on her for years suddenly didn’t need her anymore. They had something cheaper. Something that never slept, never negotiated rates, never needed a deadline extension.
She didn’t see it coming. That’s what haunts me most.
She did everything right
Sarah had done everything right. She specialized. She built long-term client relationships. She delivered quality work for over a decade. She was good — genuinely good.
But the market didn’t care about any of that.
AI translation tools had quietly gotten good enough. Not perfect. But good enough that clients stopped caring about the gap between “good enough” and “exceptional.” The economics were too obvious to ignore.
Within months, her income collapsed.
She’s lucky, and she knows it. Her husband has a stable job. They’re managing — though “managing” now means budgeting carefully in ways they never had to before. Their five-year-old doesn’t know anything is different. They’re making sure of that.
Sarah has since started an Ayurveda course. She says she’s finding something in it she never expected — a sense of purpose that her translation work, as skilled as she was, never quite gave her. Maybe this is where she was supposed to end up.
But here’s the thing I keep thinking: it’s easy to find a silver lining when you have a safety net. What about the single mother who was the only income in her household? The translator whose partner’s salary isn’t enough? The freelancer who spent years building expertise that evaporated in eighteen months?
This isn’t speculation anymore. The data is here.
In March 2026, Anthropic — the AI company behind Claude — published a landmark research paper: Labor market impacts of AI: A new measure and early evidence by Maxim Massenkoff and Peter McCrory.
What makes this paper different from the usual AI-and-jobs think pieces? It uses real usage data from millions of Claude conversations — not just theoretical models of what AI could do, but what it’s actually doing in professional settings right now.
The findings are sobering.
The most exposed occupations, ranked by how much of their actual daily work AI is already performing:
- Computer programmers — 74.5% of tasks already covered by AI
- Customer service representatives — 70.1%
- Data entry workers — 67.1%
- Medical records specialists — 66.7%
- Market research analysts — 64.8%
Translation work — Sarah’s field — falls squarely in the categories the paper flags as high-exposure: language-based, pattern-driven, and now firmly within the capabilities of large language models. The paper didn’t just predict this. It measured it happening.
Why I built this tool
What happened to Sarah is what happens when disruption arrives without warning. Not because the signals weren’t there — they were — but because no one had put them somewhere accessible, honest, and easy to understand.
The Oxford Martin School study by Frey and Osborne has existed since 2013. It modelled the automation probability of 758 occupations. Translation work scored a 38% probability — significant, but not alarming enough to act on in 2013. A decade later, the actual disruption arrived faster than the models predicted.
I built ismyjobreplaceable.com because I didn’t want anyone else to be surprised the way Sarah was. Not because knowing changes the outcome — sometimes it doesn’t — but because knowing gives you time. Time to upskill. Time to diversify. Time to make choices before they’re made for you.
The tool isn’t a prediction. It’s a map. And maps don’t guarantee you won’t get lost — they just mean you won’t be lost without knowing it.
What Sarah says now
I asked Sarah if she wanted me to write this. She thought about it for a week, then said yes. Her reasoning: “If one person reads this and starts preparing earlier than I did, it’s worth it.”
She’s not bitter. She’s practical in the way that people who’ve been through something difficult often are. She says the hardest part wasn’t the money — it was the identity. Fifteen years of being “a translator.” Knowing where you fit. And then not knowing anymore.
She’s rebuilding. Slowly, and on her own terms. The Ayurveda work is becoming something real. She’s good at it in a way that has nothing to do with being efficient or scalable or cost-effective. It requires presence. It requires a human being in the room.
For now, that’s enough protection.
Check your own risk
Don’t wait to find out the hard way.
Enter your occupation and see your AI automation risk score, based on Oxford Martin School + Anthropic Economic Index — 758 occupations.
Check my job risk →