AI Isn't Making Marketing Sloppy. Perhaps We're Blaming the Wrong Thing.

I came across a CNN article recently about restaurants using AI generated food images on their menus. Some of the examples have been mocked online and labelled “AI slop”, and I can understand why.

Some looked strange. Some looked far too perfect. Others barely resembled food you would expect to receive when your order eventually arrives at the table.

If a restaurant is selling an actual dish, I believe customers should be able to see what they are genuinely ordering. If AI creates a spectacular plate of food that the kitchen could never reproduce, that is not simply a creative decision anymore. It can become misleading.

But something else about the conversation bothered me.

There seems to be a growing tendency to see something poorly made with AI and conclude that the business is lazy because it used AI.

I don't think that's entirely fair.

Because bad marketing existed long before ChatGPT did.

We had terrible stock photographs. Badly Photoshopped advertisements. Awkward product photography. Generic captions. Menus with food photographs that looked nothing like what eventually arrived. Designers could produce bad designs with Photoshop, just as photographers could take bad photographs with very expensive cameras.

We didn't blame Photoshop.

We didn't blame the camera.

We blamed the work.

So perhaps the question shouldn't simply be whether AI was used.

Perhaps the question should be: Who decided this was good enough to publish?

AI has made creating easier. It hasn't made creating something good easier.

I use AI almost every day in my work.

I use it for research, brainstorming, writing, visual concepts, image generation, planning and sometimes simply to explore an idea that is still half formed in my head.

And yes, it makes many things much faster.

But faster doesn't necessarily mean better.

There is this impression that using AI means typing a sentence into the ai platform, pressing enter, taking whatever appears and finishing your work five minutes later.

Anyone who works seriously with AI will know that isn't always what happens.

Sometimes I spend more time telling AI what I don't want than what I do want.

An image looks great, except it has changed the product. A piece of furniture suddenly has the wrong legs. The food looks beautiful, but it isn't the dish being sold. A paragraph is technically well written but sounds nothing like the brand. Something looks polished at first glance, but when you actually look at it properly, it makes absolutely no sense.

So you change the instructions.

Generate again.

Correct it.

Rewrite something.

Reject another version.

Try another direction.

And eventually you get something useful.

But here is the important part.

You need to know what is wrong before you can tell AI to fix it.

That is where I think much of the conversation about AI misses the point.

Prompting isn't the only skill. Judgement is.

People talk a lot about learning how to prompt AI.

Prompts certainly matter. Giving clear instructions usually produces better results.

But knowing how to write a clever prompt doesn't automatically make someone a marketer, designer, photographer, copywriter or strategist.

Before I ask AI to create something for a business, I still need to understand the business.

Who are we talking to? What are we trying to achieve? What should this brand sound like? What would its customers respond to? What information must be accurate? What should never be exaggerated? Does this visual represent the actual product? Does this idea make sense outside my computer screen?

AI can help me answer some of those questions.

But I shouldn't outsource all of those decisions to it.

If AI produces ten campaign ideas in thirty seconds, somebody still has to recognise which one is worth pursuing.

If it writes ten headlines, somebody has to know which one actually sounds right.

If it generates a beautiful image, somebody has to notice that the image is beautiful but completely wrong for the business.

That somebody is still a human.

And sometimes the right decision is NOT to use AI.

This is particularly important when we are talking about food, products, people, places and events.

If I am marketing a restaurant, there are times when I want the actual dish photographed.

Not an AI interpretation of it.

The actual dish.

If I am marketing furniture, customers need to see the actual furniture they are considering buying.

If I am covering an event, I want photographs and videos of what actually happened. The expressions, interactions and moments that took place cannot simply be invented afterwards and presented as reality.

This is where I think we need to separate documenting reality from creating visual communication.

They are not the same thing.

AI can be brilliant for the second.

I can use it to develop a campaign concept, visualise an idea, create a background, explore different compositions, build a storyboard or produce something imaginative that would have been extremely expensive or difficult to create physically.

But if the purpose of an image is to show customers what something actually looks like, authenticity matters.

Knowing when to use AI may eventually become just as important as knowing how to use it.

So are photographers and videographers going to disappear?

I don't believe so.

Their work will change. It already is.

But a good photographer isn't valuable simply because he or she owns a camera and knows how to press the shutter button.

A good photographer sees things other people don't. They understand light, composition, timing and emotion. They know when something is worth capturing. A good videographer understands movement, pacing and storytelling. They can walk into an event and recognise the moments that will matter later.

Those are human decisions.

AI may become another part of their workflow. Editing may become faster. Storyboards may become easier. Ideas can be visualised before a shoot. Things that once took hours may eventually take minutes.

But making someone's job more efficient is very different from making that person unnecessary.

We have seen this before.

Canva didn't make everyone a designer.

Canva made design accessible to millions of people and that was a wonderful thing!

Suddenly, a small business owner who could never afford design software or a professional designer could create a decent poster, presentation or social media post.

But giving millions of people access to Canva didn't suddenly create millions of brilliant designers.

Smartphones put remarkably capable cameras into practically everyone's pockets.

That didn't turn everyone into a professional photographer.

Social media gave every business the ability to publish content whenever it wanted.

That certainly didn't make every business good at marketing.

AI is another enormous leap in accessibility. It gives more people the ability to write, design, generate, experiment and create things that previously required much more time, money or technical knowledge.

And I think that's exciting but accessibility to a tool and mastery of a craft are two very different things.

In fact, AI may make human judgement MORE valuable.

For years, creating was often the difficult part. You needed the software. You needed certain technical skills. You needed time. Sometimes you needed a fairly substantial budget.

AI is removing many of those barriers.

Today, someone can ask for twenty headlines, ten campaign concepts and several visual directions before finishing their morning coffee.

So what happens when generating ideas becomes incredibly cheap and incredibly fast?

Choosing becomes more important.

If everyone can generate one hundred ideas, the valuable skill isn't necessarily producing idea number 101.

It is recognising which of those one hundred ideas deserves to exist.

It is knowing what to keep, what to change and what to throw away.

It is understanding why something feels wrong even when it looks technically perfect.

It is knowing your customer well enough to realise that the cleverest idea in the room isn't necessarily the one that will work.

And there is another word for much of that.

Taste.

Taste is difficult to automate because it isn't simply about whether something is correct.

It comes from experience, exposure, instinct, understanding people and making countless decisions over time about what works and what doesn't.

Perhaps AI isn't replacing creativity. It is exposing it.

When people say AI is going to take over the world and replace everyone's jobs, I don't completely agree.

Will it change jobs? Absolutely.

Will some tasks disappear? Almost certainly.

Will one person eventually be able to accomplish things that previously required several people? In some situations, yes.

Businesses and creatives shouldn't ignore that, but I also don't think the future is as simple as humans versus AI. The more interesting divide may eventually be between people who simply use AI and people who know how to think with AI.

One uses it because it can produce something quickly. The other knows when to use it, what to ask from it, what to question, what to change and, importantly, when to ignore it completely; and that brings me back to those AI generated restaurant images.

If the food looks nothing like what the restaurant serves, I don't think the biggest problem is that an AI image generator exists.

Somebody looked at that image.

Somebody approved it.

Somebody decided it represented the business.

The AI generated it.

The human made the decision.

So perhaps we should stop asking whether using AI makes a company sloppy.

The better question is whether we are still thinking critically about the things AI helps us create. Because these tools are only going to become more powerful and more accessible from here.

Having access to AI won't be the advantage for very long. Eventually, almost everyone will have access to similar tools.

Knowing what to do with them will matter.

Knowing when not to use them will matter too.

And having the tool still doesn't automatically give you the idea.

It certainly doesn't give you taste.