Grammar and spelling errors. Tone missteps. Syntax flubs. Plagiarized copy. And flat-out fibs.
The reach of artificial intelligence (AI) tools is already vast, creating a slate of upsides for brands and agencies—jump-starting concept creation, enabling deep personalization, accelerating speed to market, and executing simple grammar checks, among others. But its downsides can’t be glossed over, especially when it comes to creating messaging that builds brand devotion among loyal customers.
Brands that rely heavily on AI without careful human oversight risk distributing flawed, misguided, erroneous, offensive, and legally questionable content. The Lacek Group’s seasoned copyediting team recently gathered to discuss how AI affects copy, reflect on how it’s changed their approach, and consider what the future of copyediting looks like.
LIZZIE DANIELS: When Lacek creates copy for our clients, it needs to feel like it’s coming from humans, not machines. Consumers want to feel valued. We’ve all seen the many studies: When customers feel a real, emotional connection to a brand, that’s when they’re the most loyal. Brands can easily damage or lose that emotional connection with mechanical, long-winded, or tone-challenged messaging that AI tools often create.
KARIN MILLER: Exactly. We know that customization and personalization are key building blocks to attract and retain loyal customers. AI without thorough human oversight can create copy that’s not in tune with the brand or that sounds like a Mad Lib, which completely derails those efforts.
KAREN BUTAK: I’d argue that a failed customization attempt can be off-putting, even insulting, to the customer. And that can distance them from the brand rather than inspire any sort of devotion. Personalization misses send a clear message that a brand is inauthentic and doesn’t really care about the recipient. That damages the relationship and stands in the way of business goals.
WENDY WECKWERTH: Yes, AI gone wrong can really undercut the brand voice and alienate the reader.
KARIN: People are savvy—they usually know when something is AI generated. In fact, a recent NielsenIQ study found customers understand when ads are developed using AI, and they find them annoying, boring, and confusing. AI copy can sound disjointed and often doesn’t have a consistent tone. For example, a headline that doesn’t align with a subhead, which doesn’t connect with the body copy—it’s as if three people wrote the various parts.
LIZZIE: We’ve all been copyediting for a long time—each of us has been with Lacek for more than a decade, some of us more than 20 years. We know the brand voices of our clients and this industry really well. It’s not surprising that AI doesn’t understand and can’t apply the nuances we’ve internalized. That’s at least part of the reason copy from an AI prompt often sounds off or is just plain wrong.
WENDY: For sure. One clear giveaway that language was created with AI is repetition. That’s something copywriters avoid, and copyeditors would suggest changing.
LIZZIE: Yes, absolutely. AI repeats words, phrases, even whole sentences.
WENDY: AI copy also seems to rely ridiculously hard on creating articles using lists—the “listicle” format. Don’t get me wrong—lists can provide a helpful structure for complex ideas and make copy more scannable. Both are valuable attributes. But not all content lends itself to a list, particularly when the goal is a warm, personal tone.
KAREN: And the nature of large language models (LLMs) means that without human intervention by writers and copyeditors, there’s a sameness to AI copy. The copy may be fine on its own, but the more companies use AI, the more people notice its patterns.
WENDY: I recently discovered a fantastic term: breviloquent, which means being brief and eloquent simultaneously. When I read its definition, my first thought was, “Now, that’s something AI isn’t doing well.” At least so far.
LIZZIE: So true! Great word.
KARIN: Coming up with concise, articulate, and evocative copy isn’t easy—it takes more than a computer and an AI prompt. It absolutely requires a writer in collaboration with an editor, plus the relevant context, a clear strategy, and an understanding of tone.
KAREN: Sometimes readers may not be able to describe what’s wrong in AI copy, but they know something’s off. Language processing is complex. That means accuracy, consistency, correctness, and all those other things copyeditors assess help make absorbing information easier for readers.
LIZZIE: Such a good point. Even subtle miscues matter. Copy can inspire devotion or lose loyalty.
WENDY: Right, a copy problem could be a phrase that’s unintentionally insulting or offensive. Or it could be an error or a typo. But sometimes a problem is more subtle. Human oversight—in this case, a copyeditor—helps ensure gaffes (both obvious and subconscious) are removed from customer-facing materials.
KAREN: I saw something not so subliminal recently. AI was asked to summarize an article, but it missed a contextual not. As a result, the AI-produced summary reflected the exact opposite of what the brand wanted to communicate. Fortunately, copyeditors incorporate fact-checking as a matter of course, so those errors were fixed. But without human oversight, something like that could go out to customers and damage the brand.
KARIN: That’s exactly the kind of thing copyeditors look for and catch, which can save clients a lot of headaches and money.
LIZZIE: Yeah, fact-checking is increasingly important. Much of what’s found on the internet, which AI scrapes to create its responses, is wrong or even fake. So the onus is on humans to make sure new copy isn’t just recirculating those inaccuracies. Instead of assuming something is factual and quickly confirming it, fact-checking in the age of AI demands a deeper dive.
KAREN: Right. We used to check, for example, the dates of a festival and the spelling of the festival name. Now we need to verify that the festival actually exists!
LIZZIE: And it’s not enough to just click a link, confirm that festival name, and move on—because that too could be fake. Finding a trustworthy source is crucial.
WENDY: Literature scholars have long discussed how authors subconsciously—sometimes consciously—“borrow” from previous works. The idea being that everything is built on what came before it. Nothing is truly original. With AI, this concept is amplified.
KARIN: I reviewed some AI copy recently with the final line: “Because you’re worth it.” That’s a well-known L’Oréal tagline, but the copy wasn’t for L’Oréal. Using a famous brand’s slogan is distracting at best. At worst, it could have serious copyright implications.
KAREN: And that sort of misstep is likely to come up regularly with the LLMs that fuel AI responses. They’re basically autofill on steroids. Trademarked phrases pop up because AI tools are regurgitating words that often go together.
WENDY: That’s a great segue into another AI offense: the dominance of clichés.
LIZZIE: Unlike copyeditors, AI loves using clichés.
KARIN: Yes! And if no one intervenes, customers get stale copy that doesn’t sound original, authentic, or genuine. No brand that truly cares about its customers wants that.
WENDY: Another thing copyeditors always look for—and need to hone in on even more with AI copy—are the problematic -isms: racism, sexism, ageism, etc. Again, because AI bases its copy on what already exists online, some of the uglier ideas people have shared can appear in AI-produced copy.
KAREN: Agreed! And AI bias can be more subtle too. A while back I worked on a presentation that used AI-developed personas. It was pretty clear the prompt had asked AI to create broadly representative personas—and, superficially, it achieved that. But the personas still reflected ingrained cultural stereotypes: a Black man and an Asian woman, not the other way around. And the Asian woman worked in finance. Plus, more than one persona referenced having a family, but the woman mentioned work–life balance was super important to her, while the man reported valuing family time but said sometimes work came first.
WENDY: Obviously, if it were up to us, everything would be reviewed by a copyeditor before being distributed. (We’re biased that way!) But it’s worth discussing the benefits of AI from a copyeditor’s perspective. AI helps our agency deliver a wide breadth of ideas to clients faster. How does it support us as copyeditors or enhance our role?
KARIN: Well, it’s clear how AI can be effective for copywriters, helping them come up with a broad range of ideas to build on. For marketing copyeditors like us, I don’t know exactly how we’d apply AI.
KAREN: I’m sure AI is helpful for book editors—simply because of the scope of a book project. But for shorter lengths of copy, like what we review, I don’t think AI contributes much more than spelling and grammar checks—and even then, it can be unreliable.
LIZZIE: Copyediting often calls for rewriting, and AI can be helpful with that. Sometimes an AI prompt can return a word, phrase, or description you might not have thought of.
WENDY: But it sounds like we’re saying copyeditors are even more necessary in the age of AI.
KAREN: I think so! If we want to create emotional connections with consumers and inspire a sense of belonging, AI shouldn’t be detectable in marketing copy. That’s why copyeditors are so important right now.
LIZZIE: AI isn’t going away. We need to find its sweet spot: using AI in cooperation with humans, including copyeditors. AI is helping marketers scale highly personalized content—among other wins. But it’s important to remember that copyeditors are a key step in making AI-assisted content feasible, trustworthy, and engaging.
For more on how copyediting can strengthen marketing efforts, see:
Your Copyeditor Is Out to Shame You (and Other Copyediting Myths)
Karen Butak, Lizzie Daniels, Karin Miller, and Wendy Weckwerth comprise The Lacek Group’s editorial team. For more than 30 years, The Lacek Group has been perfecting the art and algorithms of brand devotion, helping world-class brands identify their highest-potential customers, engage them across channels throughout their life cycles, personalize each relationship for optimal long-term results, and measure the true effectiveness of those efforts.
.png)
