Are Content Marketers Actually Using AI the Right Way?

A while back, I was scrolling through a survey, and when I finished it, I just sat there, stunned for a second.

Not because it shocked me. It was that "oh, so that's how everyone's doing it" moment of recognition — followed immediately by a quiet "wait, something's off here."

The survey came from Orbit Media. Every year they run a questionnaire for the content marketing industry, and the 2025 round pulled in nearly 800 responses. One whole section was dedicated to AI: how do you use it, where do you use it, what worries you, how well does it work.

I went through the numbers, and one thing really caught my eye.

95% of marketers are using it — but not many are using it well

Let's start with one number.

In 2023, 65% of content marketers used AI. In 2024, 83%. In 2025, 95%.

Think about what that means.

Out of every twenty marketers, only one is still holding out. AI's penetration has moved faster than the smartphone ever did. Phones took seven years to reach "everyone has one" status; AI took less than three to basically own this industry.

That's fast.

But the numbers that follow aren't as pretty.

AI's most common job has become "finding flaws"

Orbit Media listed nine options and asked respondents to check off what they use AI for.

In 2023 and 2024, the number one answer was always "finding topics, generating ideas." Makes sense — AI has read more than anyone, so asking it to help you brainstorm is the first thing anyone would think of.

But in 2025, the top spot changed hands.

"Having AI help me revise my drafts" became the most popular use.

Two out of three marketers have now slotted AI into their writing process — and its main role isn't writing, it's revising. You drop in a draft and ask it to find the flaws, suggest changes, rephrase things. Tools like Grammarly, or just using ChatGPT as your reviewer. Both roads end up at the same place.

I sat with that for a minute, and it made sense.

Why? Because when it comes to revising, AI really is faster than humans, and it's less likely to make a catastrophic error. Ask it to write a whole piece from nothing, and it'll make things up, drift off-topic, and invent facts with a straight face. But hand it a draft that's already in shape and ask where it's bloated, where it wanders, where it could hit harder — that's a job it does reasonably well.

Put simply: AI as an editor is more reliable than AI as a writer.

But some people insist on making it the writer

There's another option in the survey: "using AI to write the entire article."

What's the share?

One in ten.

And it's doubling every year. Sounds like real momentum, right?

But I'm going to throw cold water on that. This use is the least popular of the nine options, and the one with the worst results. More on that when we get to the results section.

Why don't more people do it?

Because a genuinely good piece of content is far more than "just putting words on the page." It needs original data, real human opinions, quotes from interviews, internal links that only you would know. AI can't give you a single one of those.

AI can write you ten thousand words. Whether they're any good is a different question.

There's an old copywriter named Eugene Schwartz who said something I really love: "Copy isn't written. It's assembled."

Assembled from insight, from experience, from your actual judgment on the topic. No matter how smart the model gets, it can't assemble those.

What are marketers actually afraid of?

95% of them have reservations about AI. That proportion is absurdly high.

What exactly are they scared of? The top three are these.

First, inaccuracy. 77% flagged this. When you use AI for research and writing, it speaks with confidence, and you can't always tell which sentence is real. But here's the interesting part — if you only use it for reviewing and suggesting, you can sidestep that risk. You're not asking it "what is the fact," you're asking "what do you think of this paragraph." It's giving an opinion, not a fact, and as long as you know that, you're fine.

Second, lack of originality. 63% worry that whatever AI writes comes out sounding the same. That worry is fair. AI was trained on what already exists online, so what it produces is essentially an echo of the average. But it can help you find the topics nobody's written about yet — the "blue ocean" topics. Once you've found the gap, the angle is still on you to dig out.

Third, wrong voice. Half of marketers struggle to get AI to write in their brand's voice. I've tried this myself. I fed a custom GPT more than fifty of my own articles before it even started to barely sound like me. You thought five articles would be enough? Not even close.

As for "will I get penalized by search engines," about a quarter worry about that. My take: no need.

There's no such thing as an "AI content penalty." Your content might not rank, might not get traffic — but that's because it isn't good enough, not because it's AI-generated. What you should really worry about isn't how Google judges it. It's whether your reader, after finishing it, mutters "what the hell is this."

Sure, it's faster. But does faster actually help?

This next part is, to me, the most gut-punching section in the whole survey.

Every year Orbit Media tracks "how many hours does it take to write one article, on average." For nearly a decade, that number kept climbing — all the way to 4 hours 10 minutes in 2022.

And then it dropped.

In 2025: 3 hours 25 minutes.

Break it down by AI users vs. non-users, and the gap widens: users average 3 hours 24 minutes, non-users 3 hours 48 minutes. Roughly a 10% difference.

Faster. Definitely faster.

But here's the question: is the point of marketing to be fast?

Think about it. If what you write gets no readers, no conversions, no memory, then what's the point of being 10% faster? You've just become more efficient at producing things nobody wants.

The real question: which use actually correlates with good results?

One question in the survey was: "How effective is your content marketing?" Four answers to choose from: very effective, somewhat, not really, no idea.

20% chose "very effective." That 20% is the baseline.

Then the researchers cross-referenced every AI use against that baseline, to see which uses positively correlated with strong results.

The findings are worth sitting with.

Marketers who don't touch AI reported "strong results" at the lowest rate — only 14%. Six points below the baseline. These are either the old-school holdouts who insist on writing by hand, or people who haven't figured out how to use it yet. But the data is right there: skipping AI entirely does cost you.

Marketers who use AI to write the entire article aren't doing much better. Also below baseline. If your plan is to flood the zone with AI-written content and win on volume, the data doesn't back that up. You think you're saving time; you're actually trading away results.

So which uses beat the baseline?

Using AI to find topics, generate ideas, write headlines — these "early-stage" uses correlate with strong results. Number one was "generating ideas": 23% of those marketers reported strong results, three points above the baseline.

See the pattern?

The more AI helps you think, the more useful it is. The more it writes for you, the more it gets in your way.

Read that one three times.

So how should you actually use it?

After I digested this survey, one line was left in my head.

Don't let AI do your work for you. Let it help you do your work.

Do it for you, and it turns you into a cog on an assembly line, tightening screws. Help you do it, and it becomes that partner who'll brainstorm with you at 2 a.m. and never judge your bad ideas.

Sonia Simone, a content marketing veteran, once said something sharp:

"Don't take shortcuts. Shortcuts take too long."

When I read that line, I froze for two seconds, then laughed.

Was she being sarcastic? No. She meant it: the shortcuts you think you're taking all eventually loop you back around. You skip the thinking, skip the polishing, skip the painful part that's actually what makes the content shine. You think you saved time. What you lost was the result.

This survey's data backs that up, exactly.

AI hasn't made marketing easier. It just found a new way to ask you the same old question.

Are you actually thinking?

People who think clearly: AI is their amplifier. People who don't: AI is their accelerator — it just helps them produce mediocrity faster.

As for me, when I finish a draft now, I still hand it to AI and ask where it's bloated, where it doesn't land.

But that "wait, something's off here" feeling from the very beginning — I've figured out what it was.

What's off isn't AI. What's off is the illusion that "now that I have AI, I don't have to think anymore."

And this survey proves exactly that.