Can AI replace a content strategist?

It can replace the execution part: first drafts, structuring an article, headline suggestions. It can't decide what's worth saying, for whom, and at what moment, because that requires understanding culture and brand, not just language.

Why does cultural empathy matter if AI can already write good copy?

Because grammatically correct text isn't the same as appropriate text. AI doesn't feel the Romanian context, the social moment, or the right tone for a sensitive category. The difference between a correct ad and one that resonates sits exactly there.

What should a marketing team do about AI right now?

Clearly separate execution, which automates fast, from strategic judgment, which stays human. And put a review process in place for anything generated or automatically optimized before it reaches the public.

AI in Marketing: The Job Nobody Sees

A few months ago I came across a recent Harvard study that finally asked a direct question: how morally wrong would it be if AI took over a given job. Researchers asked 2,357 people to rate 940 occupations on a scale from 1 to 7. Search marketing strategist scored 2.31, one of the lowest in the entire study.

That didn't surprise me. Honestly, I think most of my colleagues in content and PPC would give the same answer if you asked them directly. The public doesn't see marketing as a job worth defending morally when it comes to AI. We're not doctors, we're not teachers, we're not pilots. We're the people who make ads, and if an algorithm can do that faster and cheaper, who cares.

But that's exactly the problem with this score. It tells you something about public perception, not about what the job actually involves. After a good number of years in content and brand strategy, I can say honestly: the visible part of marketing, the final ad, the copy, the image, is maybe 20% of the work. The rest is judgment, cultural context, empathy toward the audience, things that never show up in a CTR or ROAS report.

Why the low score doesn't surprise us, but misleads us

The same Harvard study shows something even more relevant to us: opposition to AI taking over jobs generally doesn't come from ethical reasons. It comes from doubt that AI can actually do the job as well as a person. When people believe a job requires judgment a machine doesn't have, moral resistance shows up. When people believe the machine does just as well, that resistance disappears, whether or not it's actually right.

The public believes an AI can write an ad. And it's partly right, ChatGPT can produce a Facebook Ads text in ten seconds, complete with headline, body copy and call to action. Except writing an ad and deciding which ad deserves to be written, for which audience, at what point in the funnel and in what tone, are two completely different things. The first is execution. The second is strategy. The public confuses the two because from the outside they look the same: a piece of text shows up on a screen.

What a good campaign doesn't show

I've worked on campaigns where the most important decision wasn't in the brief, wasn't in the Google Analytics 4 data, wasn't written down anywhere. It was an instinct: that a certain joking tone wouldn't work for this particular brand, even though it performed well for a competitor. That a Black Friday campaign needed to be pushed back a day because the Romanian audience is tired of fake discounts and reacts the opposite way. That a visual that looked perfect on paper didn't work because it didn't match what the person scrolling actually felt at that hour of the day.

These are cultural judgments, not calculations. An AI can tell you what has worked statistically so far. It can't tell you what your Romanian audience will feel seeing a joke about inflation in the middle of a crisis, or watching a big brand try to sound "like us" without having earned that right. That empathy toward the audience, reading the moment correctly, is the job Harvard scored at 2.31, precisely because it doesn't show from the outside.

An algorithm optimizes for what already happened. A strategist decides what's worth trying.

That's why I think the conversation about a marketing team ready for AI shouldn't be about who keeps their job, but about who learns to separate execution from judgment. Whoever does that stays relevant. Whoever confuses the two risks being replaced themselves, because all that's left is the part AI does faster anyway.

AI can write, but it can't choose

In PPC, things have changed noticeably over the last two years. Meta Ads and Google Ads now optimize budget, bidding and targeting almost on their own. A specialist who only set up campaigns manually, without interpreting the results, really has lost that part of the job, because the algorithm does it better. But the role hasn't disappeared, it has moved. I wrote about this at length because I saw it directly in our own team: the PPC role has transformed from setting up campaigns into auditing the decisions the algorithm makes.

Someone still has to ask why the budget suddenly shifts toward the wrong audience, or why a creative that performed well a week ago "fatigued" out of nowhere. Automation took over execution, but someone still answers for the result in front of the client. And when AI gets it wrong, which happens often, especially with ads that carry cultural nuance or local humor, someone has to catch the mistake before it goes live. I've seen enough cases where nobody checked, and the result was embarrassing for the brand: AI gets ads wrong and nobody answers for exactly that reason, everyone assumed "the AI knows better."

30% now, 58% soon, but something still stays

The Harvard study also states that around 30% of occupations could be fully automated right now, with public approval. With an AI more capable than today's, that number climbs to 58%. Those are big numbers, and I think most of the execution tasks in marketing, the repetitive ones, the ones that can be clearly measured, fit into that category without much argument. Reporting, basic segmentation, first-draft copy, bid optimization, all of that will be almost fully automated, and that's fine.

But the same study shows that only 12% of occupations generate strong moral resistance no matter how capable AI becomes. Marketing, obviously, isn't in that 12%. Nobody is going to feel morally outraged if an algorithm writes the text of an ad. But that doesn't mean the whole job disappears, it just means the visible, easily measured part gets automated first. What's left, brand judgment, cultural empathy, strategic decisions, is exactly the part no public perception study sees, because it never shows up separately on anyone's resume.

What to do with this

I don't think the answer is convincing the public that our job deserves a higher moral score. That's a losing fight, and honestly, it doesn't matter. I think the answer is being honest with ourselves, internally, about which part of our work actually deserves defending and which part can be handed to a tool. A content or PPC team that spends its time on manual reporting instead of brand strategy loses exactly the part that makes it irreplaceable.

My concrete recommendation for any marketing team right now, Romanian or otherwise: take the time freed up by automation, from reporting, from first-draft copy, from campaign setup, and put it directly toward the things the client can see but an algorithm can't. Audience research done with real people, not just Google Analytics 4 segments. Testing tone and humor before launch. A clear review process for any ad generated or optimized by AI, before it goes live. None of that gets automated any time soon, and that's where the difference shows between an agency that just runs campaigns and one that actually understands its client's brand.