Marketers, Is ChatGPT a Help or a Hazard?

A while ago, I opened ChatGPT and asked it to write a product intro for me.

Less than thirty seconds later, three paragraphs came back. I read them and did a double take.

The writing was mediocre. But the thing was — it was usable.

My first thought: shouldn't the marketers who make a living with words be getting spooked?

Let's start with the upside: fast. Genuinely fast.

Two months after ChatGPT launched, it crossed a hundred million users. No other product has ever moved that fast.

Marketing, as a trade, is a content grinder by nature. Ad copy, tweets, blog posts, case studies, sales pages, white papers — produced every single day. A case study used to mean hours: interviews, meetings, drafting, revisions, design. Now? Seconds, and the first draft is sitting right in front of you.

I asked around among my marketing friends. Eugene Cheng runs a consultancy in Singapore, and his one-liner: AI sped his workflow up by several hours.

Ashley McAlpin leads a marketing team, and she said she used to spend all her time writing baseline content — now she can finally invest it in insights that actually matter.

David Gómez-Rosando runs a design firm, and he put it even more bluntly: for non-creative copy, he basically doesn't need to hire writers anymore.

Listen to those three — different work, different scale, and they're all saying the same thing: time saved, costs down, output up.

Pretty great, right?

But behind every good thing, there's always a "but."

So I kept asking them: what are you worried about?

Turns out, they're worried about quite a lot.

Accuracy.

ChatGPT will make things up with complete confidence. It can pin a nonexistent article on a real journalist and fabricate a string of references that look perfectly plausible. One respondent, speaking anonymously, said something that stuck with me: "AI is infringing on copyright, and I don't want to support that."

Which leads straight into the second problem: copyright.

Where does the data for training large models come from? Scraped off the internet. Inside it is copyrighted text and images, and the original creators were never asked, never paid a cent. You take what AI generates and put it to use — what if it runs afoul of something? That risk, you carry yourself.

And another: the editing adds up.

Several people told me they'd assumed AI output would be ready to ship. Turns out, you still have to review it and revise it yourself, and the workload doesn't shrink much. Put bluntly, text coughed up by a machine out of a black box is a long, long way from "ready to publish."

What really unsettled me were a few other things.

All of the above is just the surface.

Think deeper.

Will creativity atrophy?

Youssef Hani, a marketer in Cairo, said something I found especially precise: what we really should be wary of is how quickly we can now mass-produce content, regardless of whether that content is worth producing.

He used a word: frequency. When production becomes too easy, the barrier disappears, and the garbage rolls in. If managers fixate only on the speed and savings in front of them and throw out quality and thought, they're getting it exactly backwards.

A muscle you don't use will atrophy. Creativity, probably, works the same way.

Creators aren't getting paid.

Some people figure, "I can't afford a designer, so I'll use AI to fill in — that's fine." And they're not wrong. But if you're at a company of any real size, don't rush to fire your design team. Their eye, the experience they've accumulated — none of that is something any model can replace. And the artists whose work was scraped to train these models still haven't seen a cent of compensation.

Language and culture will get swallowed.

The data large models are trained on is overwhelmingly English, overwhelmingly Western. The "good articles" they generate are essentially restatements of that training set. Over time, non-English, non-Western voices will only get quieter. The digital divide will only widen.

The electric bill.

Training GPT-3 once consumed 1.287 million kWh of electricity. That's equivalent to a year of power for 120 American households, and it released 502 tons of carbon. Crypto got hammered for its power appetite, and AI's hunger is bigger still.

And the most lethal one: confidential data leaks.

OpenAI's terms spell it out: content you enter through the web version may be used to optimize the model. In other words, there's no guarantee that every line you type stays inside your company.

Samsung learned this the hard way. Employees pasted their own source code into ChatGPT. The code walked out the door.

So, when something goes wrong, who's responsible?

By the time I asked the question, I already had an answer in mind.

The one responsible is always the person holding the tool.

Gary Marcus, a veteran of AI research, doesn't mince words: marketers absolutely have to take responsibility for what they produce with AI, and they have to actively push back against that "nobody's checking, I'll just let AI churn out a piece" impulse.

Katrina Ingram, who works on ethical AI, makes a fair point too: professional marketing and advertising already have industry standards, platforms have legal obligations, and different categories have different compliance requirements. These existing mechanisms can be applied straight to AI-generated content.

The legal side is moving too. The US Congress is already pushing AI legislation, Europe is working on the EU AI Act, and China moved a step earlier with its own regulations. This time, lawmakers have, surprisingly, not fallen far behind the technology.

One last thing, from the heart.

The tools will keep getting stronger, the people using them will keep growing in number, and the abuse won't be small. This back-and-forth tug-of-war isn't going to stop anytime soon.

There's a line I keep coming back to: artificial intelligence has already put out to sea — the question is, a ship still needs a human captain.

You're the marketer. You're the captain. Don't hand the wheel to the machine.

That job, it can't do.