I recently posted a family history compendium to this web site. When I started this part-time writing project more than fifteen years ago, artificial intelligence was dream-stuff for a handful of computer scientists. When I finished, there were three AI assistants—Claude, Gemini, and ChatGPT—each at my actual fingertips, asking at every touch how they could help me. My family history certainly needed a copy edit, so I gave that job to Claude.
A few hours and twenty dollars later, my forty thousand words of family history were free of typos and style inconsistencies and blunders like writing “favor curry” when I meant “curry favor.” And in the few seconds it took for Claude to edit each section, it threw in some fact-checking, as when I wrote in one draft about a law that Illinois governor George Ogilvie signed into law in 1969. His correct name was Richard Ogilvie.
I am careful not to let Claude do any of my actual writing. It must be very tempting to ask an AI assistant to write a paragraph on—you name it—or to recast some janky sentence, but I don’t. What I love most about the writing process is the part where you write. Why would I let any other being, sentient or not, steal that joy? I regularly set boundaries, such as by including “check only for typos” in my prompts. I find it easy to forget I’m interacting with a machine. I tell Claude “thank you” more often than I care to admit.
In the 1780s, my ancestors Christopher and Margaret Durbin moved their family from Maryland to Kentucky, presumably with the help of a wagon. Should the iron rims on their wagon wheels have ever needed repair, a blacksmith would have fixed them. In the 1940s in East St. Louis, my grandmother Ann Durbin put a numbered card in the window when the kitchen ice box needed a refill, signaling to the ice man to lug that many pounds up the stairs. Technology made the jobs of the blacksmith and the ice man obsolete by changing how humans live. With the invention of AI, technology continues to change the way we do all sorts of things. Including how we write.
Much of editing involves typo correction, style conformity, and fact-checking. In the past, human copy editors have done this essential work. But now I expect AI will handle much of it just fine. Sadly, I can only imagine how many human copy editors and fact-checkers will lose their jobs. But I cannot imagine AI ever fully replacing editors, because their contribution goes well beyond what any machine can do. Here’s why.
I was once fortunate enough that The New York Times ran an op-ed I had written, but only after then-editor (now investigative reporter) Graham Bowley helped shape it into something no machine could ever have produced. The reason? AI depends on what it can learn from what’s been written before. But as any writer who has worked with a good editor knows, human editors don’t just compare a draft manuscript to what’s been written before, but rather see it as the nascent form of something that has never existed. They are not simply conforming. They are creating.
As the emergence of AI continues, there are deeply troubling issues yet to be resolved—starting with the use of copyrighted material for the training of AI machines. And then there’s education: Is AI a fast-track to acquiring human knowledge? Or will it deny countless young people the gift of learning critical thinking? The list goes on.
I don’t know how the AI revolution will play out. But I’m pretty sure the next time I need a basic copy edit I’ll put Claude back to work.
Postscript: Yes, I had Claude edit this essay. It caught one typo and three punctuation errors, and suggested recasts of two sentences—each of which is better now.
Michael Durbin 