I Don’t Know if AI Is Coming for Your Job, But Here Is What It Is Doing To Mine

Every few days, another newsletter lands in my inbox warning that the robots are about to take everyone’s job. The copywriters are panicking, the coders are updating their resumes, and the graphic designers are looking into carpentry. I can’t tell you whether generative artificial intelligence is going to replace your tax accountant or write the next Great American Novel. But as someone who spends his days squinting at centuries-old handwriting and trying to make sense of historical economics, I can tell you what it’s doing to mine.

It isn’t taking my job. It’s turning me into an editor for a remarkably eager, wildly fast, and occasionally delusional research assistant.

For the uninitiated, historical research—particularly when dealing with 19th-century commercial manuscripts—is a slow, solitary grind. You sit in front of a monitor or in a dimly lit archive, staring at ink that bled through rag paper two hundred years ago. You decipher the erratic handwriting of Spanish merchants, piece together obscure vessel names, and reconstruct long-forgotten trade networks line by painful line. It is high-friction work.

Enter the Large Language Model.

Drop a digitized page of early 19th-century cursive into the prompt box, and within seconds, the machine spits out a full transcription and translation. The speed is staggering. What used to take an entire afternoon of paleographic decoding happens in the blink of an eye.

The catch? The machine is a confident bullshitter.

It doesn’t tell you, “I’m not quite sure about that squiggle on line four.” It simply guesses with total conviction. It will happily turn a historical merchant named Francisco de Borja Migoni into “Francisco de Paula Misoni” without batting an eye. It will read the name of a merchant vessel—say, the Tarantula—and transform it into absolute nonsense because the script was a little cramped near the margin.

If you trust it blindly, you aren’t doing history; you’re publishing machine-generated fiction.

And that is where the nature of my work has fundamentally shifted. The core labor is no longer just the manual labor of transcription. The labor is now verification. The job is cross-checking every proper noun, verifying every vessel log against independent databases, and knowing enough context to call out the algorithm when its imagination gets the better of it.

The machine brings incredible speed and unprecedented processing power to the table. But context, historical nuance, and the ultimate responsibility for truth still belong entirely to the human staring back at the screen.

Is AI coming for your job? Maybe, if your job consists entirely of tasks that require no judgment, no domain expertise, and no healthy skepticism. But if your work relies on knowing the difference between what looks right and what isright, the game isn’t over. It just has a whole new set of rules.

PS This blog was written by AI. None of my RS (real stupidity) was required