We’ve worked with several marketing contractors over the past year, and I keep having the same experience. Someone sends me a draft, I read the first few sentences, and I immediately know AI wrote most (or all) of it. There’s nothing wrong with that. I use AI constantly. We teach people how to use AI. I’d actually be concerned if someone working in marketing today wasn’t using it. What bothers me is when they just copy, paste and move on.
If your job is creating marketing content and you send me something that looks like you copied it directly out of ChatGPT, I start wondering why I’m paying you. That may sound harsh, but it’s worth thinking about. AI can already produce a perfectly competent first draft of everything. If your contribution is essentially pressing Enter, you’re making an argument that the client could eliminate the middleman.
I’m paying you to do something meaningful with the draft after AI generates it. You should know the audience, the brand, what sounds strange, and, hopefully, where the AI has produced something technically correct that is also definitely AI slop. That judgment and analysis are what people are paying you for. We can hit Enter on our own. We can also build an agent to do it. And lately, I’ve been seeing far too much marketing content where nobody appears to have done that part.
Wikipedia actually has an excellent page called “Signs of AI Writing” that catalogs many of the patterns people have started recognizing in AI-generated content. At this point, those table stakes. Anyone who writes professionally using AI should know them. But there are a few things I see over and over again that drive me particularly crazy.
1. Filler words that aren’t doing anything
AI loves adjectives and adverbs. Loves them in an unhealthy way. Two of my personal favorites are “real” and “quietly.” Everything has “real impact,” creates “real value,” delivers “real results,” or addresses a “real problem.” As opposed to what? Imaginary value?
Then there’s “quietly.” AI likes to describe how industries are quietly changing, expectations are quietly shifting, technology is quietly reshaping work, and professionals are quietly reconsidering everything. Most of these things don’t make sound in the first place.
And how about “what matters”? This is why it matters. That’s why THAT matters. Here’s what really matters. Just make the point and let the reader decide. Spoiler: it really doesn’t most of the time.
If removing a word doesn’t change the meaning of the sentence, there’s a decent chance the word doesn’t need to be there. Good editing requires noticing that.
2. Fake drama
AI is also very dramatic. “The real challenge is…” “The bigger question is…” “The uncomfortable truth is…” “The real opportunity lies in…”
The real question is, yes, pun intended: AI, are you okay? You must be bored because you’re taking an ordinary HR memo and turning it into a telenovela. My favorite is “the real question.” What question? Nobody asked a question.
Drama is one of the biggest tells of synthetic writing. Instead of making the point, the language announces that an important point is about to occur. Human writers certainly do this too, but good ones tend not to do it every fourth sentence. Sometimes the point is just the point. It doesn’t need ominous music behind it.
3. Three. Things. Everywhere.
AI loves groups of three. Three adjectives. Three examples. Three benefits. Three problems. Three bullets. Three clauses in the same sentence. Once you start noticing it, you can’t stop noticing it.
The rule of three is a legitimate writing technique. Humans have used it forever because three items often create a pleasing rhythm. The problem is frequency. When every concept arrives in a perfectly balanced group of three, writing starts to feel engineered rather than written. Sometimes there are two things worth saying. Sometimes there are four. Sometimes there’s one. Sometimes there are seven, because life is messy and humans are unruly.
Let the idea determine the structure instead of letting the structure determine the idea.
4. No contractions
This one seems small, but it bothers me constantly. People use contractions. We say “don’t,” not “do not.” We say “it’s,” not “it is.” And if we’ve got decent grammar, we say “you’re,” not “you are.”
Obviously, there are situations where more formal language is appropriate, but most marketing writing is supposed to sound like one human being communicating with another. AI often defaults to the more formal version unless it’s given instructions about voice. That’s how you end up with paragraphs full of phrases like “You do not need to completely change the way you work,” “There is no single approach that will work for every organization,” and “It is important to consider…”
Nobody talks like that for six straight paragraphs unless they’re delivering testimony to Congress. Changing “do not” to “don’t” isn’t revolutionary editing. It takes about two seconds. But those two seconds tell me someone actually read the draft.
End of rant.
There’s absolutely nothing wrong with using AI to create marketing content. Please use it. Let it give you the first draft. Let it generate ideas. Let it… pick your favorite AI-generated triad to complete this beautifully structured threesome.
Then read what it wrote. Read every word. Delete the filler. Remove the fake drama. Break up the patterns. Use contractions. Rewrite the sentences that don’t sound like your client. Add the examples the AI couldn’t know. Challenge anything that feels generic. Make decisions.
Bring some judgment to the process because AI has changed the economics of writing. A client no longer needs to pay someone simply because that person can produce competent sentences. The client can produce competent sentences too.
What they still need is someone who knows what those sentences should say. They need someone who understands their audience, notices when something sounds wrong, knows what belongs and what doesn’t, and can turn a generic draft into something worth publishing. They need someone who puts in the work and does something valuable that AI can’t.
Copying and pasting whatever AI gives you… that ain’t it, kid. (That’s a Broadway reference.)
If you’re doing that, you may want to think carefully about the message you’re sending your clients, because sooner or later one of them is going to have the same thought I keep having: Why am I paying someone else to do something I can ask AI to do myself?
For the love of all things holy, read what AI generates. Use the tool. Then do the part of the job we still need humans for.