Artificial intelligence is one of the most useful things that has happened to marketing in years.
It is also becoming one of the easiest ways to produce very bad marketing, very quickly.
Lately, more and more clients arrive with something already prepared: a social media strategy generated with ChatGPT, a list of content ideas, ready-made captions, proposed campaigns, even finished creatives.
And there is nothing inherently wrong with that.
In fact, these tools can be incredibly useful for organising thoughts, exploring ideas and creating a first draft.
The problem begins when the draft is mistaken for the strategy.
AI can give you a very convincing answer — even when you ask the wrong question
There is a simple experiment anyone can try.
Take a social media creative and show it to an AI tool.
Ask:
“Don’t you think this design is far too simple?”
There is a good chance you will receive a thoughtful explanation of why the design is too simple: perhaps it lacks hierarchy, visual interest, information or differentiation.
Now start a new conversation, upload exactly the same creative and ask:
“Don’t you think this design is far too busy?”
You may receive an equally persuasive explanation of why the design contains too much information, lacks breathing space and creates unnecessary visual noise.
Same creative.
Two opposite conclusions.
Both convincingly argued.
This is one of the most important things people need to understand about using generative AI professionally: a well-written answer is not necessarily a correct answer.
And a confident explanation is not the same thing as professional judgement.
Leading questions are especially dangerous because they often contain the conclusion before the analysis has even started.
Social media is beginning to look like a collection of unfinished brochures
You can already see the result.
Social feeds are increasingly filled with posts overloaded with text, subheadings, explanatory paragraphs, icons, arrows, statements, disclaimers and calls to action — all competing for attention inside a single graphic.
Many of them look less like social media content and more like the first draft of a corporate brochure.
Nobody knows where to look.
Nothing has enough visual weight.
And the communication fails before the audience has even had the chance to read it.
This does not mean that the information itself is useless.
Quite the opposite.
A client-generated AI draft can be an excellent starting point.
It can tell the marketer:
“This is what I want to communicate.”
“These are the points I consider important.”
“This is how I understand my audience.”
Perfect.
That is valuable input.
Then comes the part that requires experience: deciding what stays, what disappears, what becomes a headline, what belongs in the caption instead of the creative, what needs its own post, what fits the brand and what the audience is realistically going to stop scrolling for.
That is marketing.
AI can create a marketing plan. That does not make it a marketing strategist.
Of course ChatGPT can generate a marketing plan.
Give it a company description, a target audience and a goal, and within seconds it will produce objectives, content pillars, suggested platforms, campaign ideas, KPIs and a posting schedule.
Some of it may even be very good.
But producing a document called a “marketing strategy” is not the same thing as having strategic judgement.
A real strategy is shaped by context.
It is shaped by experience with the client, by previous campaign performance, budgets, internal politics, competition, audience behaviour, timing, brand positioning, commercial priorities and dozens of small details that rarely make their way into a prompt.
It also requires knowing when standard advice should be ignored.
That part is much harder to automate.
Marketers use AI too. We just use it differently.
I use AI constantly.
ChatGPT is part of my everyday workflow. So are Midjourney, Canva, CapCut and, occasionally, Gemini.
Most marketers I know use some combination of AI-powered tools.
The difference is not whether professionals use AI.
The difference is what happens after AI produces the first answer.
We edit.
We reject.
We rewrite.
We compare.
We challenge the output.
We know when something sounds right but does not fit the brand.
We recognise when a suggestion may work in theory but not on a specific platform.
We remove things.
Very often, the most valuable thing an experienced marketer does with AI-generated content is delete half of it.
AI accelerates the work.
It should not replace the thinking.
After enough exposure, AI-written copy becomes surprisingly easy to recognise
There is another side effect of the explosion of AI-generated marketing content: everything is beginning to sound strangely similar.
Certain sentence structures appear again and again.
Certain dramatic openings.
Certain transitions.
Certain polished but slightly inflated expressions.
Certain phrases that feel perfectly natural in English but oddly translated when transferred directly into Greek.
After thousands of hours working with these tools, you start recognising the pattern almost immediately.
Sometimes I can tell that a Greek caption was generated almost entirely by ChatGPT from the first few words.
That does not mean everything AI writes is bad.
Far from it.
I have personally adopted expressions I first started using more frequently through AI-assisted writing. Some are genuinely useful because they are more precise than the wording I would previously have chosen.
“Medical evaluation”, for example, often describes something broader than simply “medical examination”.
“Scientific evidence” or “scientific substantiation” can also be exactly the right wording in the right context.
The goal is not to eliminate every trace of AI.
The goal is to make sure there is still a human being making decisions.
AI does not replace expertise
The same principle applies far beyond marketing.
An AI system can give you information about an injury, help you understand possible causes and explain medical terminology.
That does not turn it into the doctor examining you.
Similarly, it can suggest campaigns, captions, visual concepts and marketing frameworks.
That does not make it the professional responsible for your brand.
There is a difference between having access to information and knowing what to do with it.
There is a difference between generating options and choosing the right one.
And there is a very big difference between producing content and building a brand.
The real danger is not AI. It is misplaced confidence.
AI is not killing marketing.
Used properly, it is making good marketers faster, more productive and more capable.
The danger begins when its output is treated as expertise simply because it is immediate, polished and confidently written.
A weak strategy can now look extremely professional.
A bad idea can arrive with ten bullet points explaining why it is brilliant.
An ineffective campaign can be produced in minutes rather than weeks.
And if businesses trust the output without applying experienced judgement, the cost will not simply be a few unattractive Instagram posts.
It can mean wasted advertising spend, diluted positioning, inconsistent communication and months of moving in the wrong direction.
Use AI.
Use it extensively.
Let it challenge you, help you organise your thinking, create alternatives, analyse information and accelerate production.
But do not confuse access to a powerful tool with mastery of the profession.
AI is an extraordinary assistant.
It is still a very poor substitute for experience.