After a Marketing Team Hands Content Production Over to AI

A while back, a friend of mine in consumer goods poured out his frustrations to me.

His company kept four copywriters, two designers, and one video editor on staff — salaries alone ran a few million a year for the content team. The boss winced at the cost. After sitting through a talk early in the year, he came back and made the call: we're going all in on AI. Three months in, output had multiplied several times over — and yes, headcount was down.

But at the mid-year review, conversions hadn't budged. Followers were still churning. The boss couldn't make sense of it.

I told him — this is genuinely interesting. And today I want to talk it through with you properly.

Let's Start With the Part That Tempts You

AI writes fast. Really fast.

How fast? A copywriter who used to agonize over a single post for the better part of a day can now get a first draft in minutes. A week's worth of content gets scheduled in a single afternoon.

I can't deny how good that is. Right?

And it isn't only speed. The thing can do the math. Feed it what a user has clicked, what they've bought, how long they've lingered, and the next email or push notification it hands you will understand that person better than you could have written yourself.

Cheaper. Faster. And it knows your user better.

Put those three on a slide — what boss wouldn't be tempted?

My friend was tempted. A lot of bosses were. So I'll say it: the upside is real. I can't let the problems I'm about to walk through become an excuse to write the whole thing off with one stroke.

But.

Every story reaches a point where the "but" shows up.

My friend's output had climbed, and conversions hadn't moved. He went back and actually read what the AI had been writing. And he saw it.

You can't find a single grammatical flaw — but the moment you finish, nothing sticks.

Every piece is correct. Every piece is error-free. And every piece reads exactly like the last. Like a batch of parts off an assembly line — the finish, the dimensions, the weight all dead on spec, but you're not going to feel anything for any one of them.

I think anyone who has actually tried AI writing knows exactly what I mean.

A Very Expensive Illusion

Why does this happen?

Think about how AI learned to write in the first place. It read an enormous volume of text, found the patterns, and now generates new text to match those patterns. In other words, what it writes is "the average, least-likely-to-be-wrong kind of article."

Average means safe.

It also means forgettable.

Drop a forgettable article into a social feed, and the user scrolls right past without feeling a thing. There's no point of view, no edge, no warmth of a real person hiding between the lines. The user doesn't dislike it — they just don't care. And the moment they don't care, they don't buy.

That's the root of my friend's flat conversions. What he bought was capacity, not connection.

The part that really stings is this: once everyone is using the same batch of tools and the same playbook to mass-produce content, the market fills up with stuff that all looks the same. Users see enough of it, they get fatigued, they start spotting "oh, this is AI again" at a glance — and they scroll away on instinct.

It's like walking into a food street where all ten shops are serving the same pre-made meals. The first bite is fine. By the third shop, you'd rather go hungry.

A Few More Pits You Should Know About Up Front

Beyond "forgettable" — the biggest pit — there are a few other things I think every team planning to use AI for content ought to be honest with itself about ahead of time.

First, it will say nonsense with a perfectly straight face. AI doesn't sort true from false; it only sorts "does this sound like it belongs." Outdated information, rumors repeated until they became "fact" — it can swallow all of it whole and deliver it with an airtight-sounding explanation. If you don't fact-check, the error lands on you. Ship one article with a glaring factual flaw, and what users remember isn't how efficient you are — it's how unreliable you are.

Second, it arrives carrying bias. The data it was trained on already has bias baked in. Gender, region, occupation — without meaning to, it can quietly write some stereotype straight into your brand copy. The moment a user catches it, the brand has to spend ten times the effort to claw the trust back.

Third, people get lazy. I've watched this happen in several teams with my own eyes. The tools are too good. The copywriter stops thinking for themselves. The designer stops coming up with their own ideas. Day by day, the thing that actually makes a team valuable — judgment, imagination, empathy for the user — atrophies a little more. The machine didn't replace you. You replaced yourself.

So How Should You Actually Use It?

At this point you might be asking: so should I just not use AI?

Of course not.

My view is plain: AI is a shovel, not the mine.

The mine is your understanding of the user. It's your judgment. It's the things living in your head that only you can say out loud. The shovel is the tool that helps you dig those things out faster and put them on the table better.

You can let AI produce the first draft — but the final draft has to have a human watching over it, shaping it until it carries the flavor only your brand has.

You can let AI do the math on user preferences — but the tone, the stance, the line that lands with people, that's your call to make.

You can use it to buy back time — but that time you saved shouldn't go to scrolling short videos. Go talk to your users. Go sharpen the take that nobody else can copy.

My friend eventually figured this out. He didn't fire the team. He moved the copywriters from "the people who write the words" to "the people who supervise the AI and set strategy." Half a year later, the capacity was still there — and the conversions slowly came back.

One Last Thing

I once read a line that I think puts it perfectly: tools amplify whatever you already bring to the table.

If you already had the insight, AI makes you fly faster. If you hadn't thought it through, AI just helps you run faster — in the wrong direction, and into more mistakes.

So don't ask what AI can write for you.

Ask yourself first: what is it you actually want to say?

Get that straight, then pick up the shovel. You'll find that what's truly valuable was never in the tool to begin with.

May your content have a human pulse — and a business behind it.