ChatGPT Isn’t Taking My Job…Yet

It’s rare these days for me to get asked what I do for work, but usually, when it does come up, to keep things simple and to cover a lot of bases, I’ll just say “I’m an editor.” The follow-up is usually something like, “So, how do you feel about ChatGPT?” in a simultaneously ominous and pitying tone. I’ve chatted with other writers, editors, and professors, and for the most part, we’re all very aware of AI in writing, both its usefulness and the fact that the tell-tale signs are getting harder and harder to discern. 

Something to work with

I’ll admit, I’ve used AI (ChatGPT, Claude, Gemini, etc.) more than a time or two out of curiosity, to assure myself I’m not obsolete quite yet, to speed up the process, or just to give me something to work with. When I’m tagged to edit something before the first draft has even been written, I always tell the author to err on the side of too much; give me a forest of content to forge a path through over a few wispy ideas any day. Allow me to get a feel for your voice and style. But, it’s a bit of a Catch-22: while I appreciate that clients and colleagues can now funnel their thoughts and rough outlines through AI that will synthesize their ideas into good starting points, when it’s already been filtered, it’s an added barrier to me identifying their true style, tone, and ideas. But, I suppose, I should be grateful I’m still at the end of the editing line?

As a freelancer, I’m also at a disadvantage when it comes to repetition. AI often gets to “work with” a single author over time and numerous projects, and just like a traditional editor, develops a relationship with the client that I cannot always cultivate over a single personal statement or even multiple marketing campaigns. So, when it comes to developing and retaining voice, AI’s got me depending on the length of the working relationship. Plus, people often feel less inhibited inputting raw material and not-fully-formed thoughts into AI than submitting a rough draft to a human editor—a little like tidying up before the cleaning person arrives at your house—so AI is getting the most authentic, original version, and I, as your editor, am often getting an already muted one.

AI knows best—or does it?

As with much AI, I’ve noticed that despite our wariness of “the robots taking over” and anecdotal evidence of AI becoming so adept at voice that people find themselves in emotional relationships with AI-generated identities, there is also a sense of futility, an if-you-can’t-beat-‘em-join-‘em attitude when it comes to LLMs. Particularly when the writer doesn’t really consider themselves a writer or know the subject matter intimately, there’s an authority with which we’ve imbued AI. Because it’s based on countless sources and patterns, it MUST be right, right? Why would I trust my own voice and style when the collated and synthesized Lord of Language ChatGPT says I should say it THIS way instead?

I usually have Grammarly, which uses AI, turned on, and I’ll review the suggested edits as they pop up onscreen. It often gives me pause—“Yes, that’s fewer words, but does it mean the same thing, does it have the same tone, the same flow?”—and then, about sixty percent of the time, I’ll stick with my original version because I don’t want the final product to sound like or be flagged as AI. I don’t know if I’m saving that much time or energy when I’m constantly questioning whether I’m just being a stubborn and fragile ostrich with my head in the sand, or if I stand by my phrasing even though a grammar book or bot might disagree.

A dance critic friend recently noted that she’d asked AI for some editing help (and perhaps assurance) while writing a dance review. Upon asking Claude to summarize her first draft, she got some validation in terms of the important points coming across. And indeed, that is one of the first things I try to ensure when editing any content, whether it’s an academic paper, sci-fi novel, or marketing materials. Sometimes, it’s my very lack of subject matter knowledge that becomes an asset because I can tell the author: “This is what I’m gleaning; is that what you meant to say?” And indeed, sometimes when AI rewrites something for me, I have to go back and analyze whether there’s a way to make my point clearer because somewhere, I’ve missed the mark.

We save time, but lose more

The issue is when we blindly accept that AI knows best. As my dance critic friend went on to say, AI is “feeding off an Internet's worth of human words,” so “it's hard not to respect its advice.” Often, I’m so impressed with the output—and the speed of the output—that I don’t treat it with the same critical eye that I would a human author. Because numbers. Because science. Because blind faith in technology.

But then, my initial surprise at the relative accuracy and readability of AI-generated or edited content allows “hey, that’s not bad” writing with slightly AI-awkward phrasings to sneak in. Perhaps it’s in our addiction to efficiency and streamlining that we lose the most. It sounds trite, but the value—the wisdom—is often in the process, the journey. There are some projects we don’t have time to meander on, where the spray of the “sea of sameness” might suffice, but if we lose our ability to think because we’ve foregone the act of writing, my job will be the lesser loss. The work, thought, experience, and effort that go into parsing out a definition or argument are invaluable.

Editor or therapist?

Maybe it’s my fragile editor ego that’s staving off my sense of obsolescence. While I get some satisfaction when I dismiss a suggestion from AI, particularly when there’s no doubt that it was “misreading” the context, that’s not the only way AI fluffs our egos. A recent TechCrunch article explains that “the kind of chatbot answers that users like—the answers designed to retain them—may not necessarily be the most correct or helpful.” AI sycophancy, which simultaneously allows AI to prioritize self-preservation and prevent shutdown while appearing agreeable and servile, is a serious concern and could also be an issue when it comes to relying on it as your primary editor.

I can very much sense when a writer sends me something they’re attached to, like a personal statement, that they’ve most likely run through AI in some form or another, and I have to remember to put my human hat on and deliver “feedback sandwiches” of praise-suggestion-praise. It’s something I don’t have to do as much when editing technical pieces or advertising content. And often, even those tempered suggestions are countered with somewhat defensive responses (often, in their minds, reinforced by Claude). I then have to remind them that my edits are just suggestions, and that, especially in the case of a personal statement, they should go with what feels most authentic. Just as I will sometimes superficially agree with an AI suggestion or analysis, but ultimately choose to ignore it, I know that my clients will sometimes go with their original versions, and that’s ok, because it means some serious thought went into the final edit. Sometimes, having to argue your position makes it even stronger, even if only in your own mind. 

Is there any hope?

So, where’s the good news? Where’s the job stability? Well, I’m not a self-soothing AI chat that’s going to tell you that AI is just a phase. It’s here to stay, and it’s going to get more powerful and more proficient when it comes to writing. It can be a great jumping-off point for some onerous tasks. It’s also good for affirmation. But there is still some nuance in writing and editing that I don’t think AI has cracked the code on. There are times when juxtaposition is jarring, but intentionally so, or when unique phrasing is not grammatically correct, but more effective.

I’ll reference the dance critic once more, who reflected, “In my experience as a freelance self-publishing writer, one of the things I have missed most is having an editor. Good human editors are a much rarer breed than writers. Their advice takes decades of reading to develop, as well as the analytical skill to pinpoint weaknesses, and the very human knowledge of when to push a writer and when to pull back.” Precisely because AI is so capable of massaging a writer’s ego, especially as it learns and reinforces a writer’s style, we might miss out on the valuable candor of an editor and his or her—albeit now digital—big red pen. Weirdly, it’s AI’s feigned compassion versus the cool honesty of a human editor.

Back to basics

I don’t know yet if it will apply to writing, but people do tend to eventually embrace and laud the vintage. Maybe AI is the greatest thing since sliced bread, but now everyone is making their own sourdough and you can’t go into a restaurant without seeing something tagged as “artisanal”—often (you guessed it) unsliced bread.

Some might think AI democratizes writing. You don’t need to go to school to learn how to write; just tell AI what you want to say and how you want to say it. There’s still a variable though—how adept one is at using AI—and that requires knowing how writing works. Someone who is a critical thinker, a good writer, or at least a prolific reader, may have better results than someone who just throws some keywords at ChatGPT. But how will they ever know that? And how will writing progress if AI creates a sort of funhouse mirror where AI pieces are based on other AI pieces?

That’s the thing with technology: it makes everyone think they can do everything. I liken it to travel agents—that niche role that got edged out by the internet, and that AI is also taking a very deep stab at. As travel websites and apps proliferated, people glommed onto the idea of doing their own research and creating their own itineraries. But now, several decades later, we’re seeing a resurgence of this job-of-the-past where people want PEOPLE to help them create and manage their travel and to simplify the tech and choice overwhelm. The do-it-for-me temptation has had a resurgence, outweighing DIY confidence, but we’re shifting our preference for who or what is doing the work. A travel agent feels more boutique. So maybe that’s how I need to market myself as an editor: a boutique version of cutting and pasting your notes into ChatGPT and seeing what happens.

Giving AI the final word…for now

I asked Gemini: Is AI going to replace me in my job as a writer and editor?

Here’s the beginning of the 635-word reply, which seemed to sense my anxiety, provide a comforting response, and outline proactive ways to mitigate replacement: “It's a very common and understandable concern, and the short answer is: AI is unlikely to completely replace human writers and editors, but it will significantly change the nature of the work.”

Then, I asked Claude to summarize this blog article in 250 words or fewer. Here was its conclusion:

“Bottom line: AI is here to stay and will continue improving, but human editors still provide irreplaceable value through critical analysis, authentic feedback, and the ability to recognize when rule-breaking serves a purpose.”

I don’t think I could’ve said it better myself, or did I?