Is AI making brands lose their cool?

1 minute ago

7 min read

I’ve been thinking about what role AI should actually have in building a brand.

For context, I’m a creative who also uses AI in my creative work. I’m interested in what it opens up, and I can still spend hours fixing an image until it feels right.

Both of those things are true.

So this isn’t a conversation I can have from a distance. I’m using these tools while questioning what they’re doing to the work, to our expectations, and to the way companies value creative people.

And something I keep coming back to is this: you can make every brand asset look better and still end up looking like everyone else.

The images are polished. The language sounds confident. There’s more content than ever.

But where is the point of view?

What are we calling “cool”?

yerba mate
cool shot
ambush

When I think about a brand being cool, I don’t immediately think about how beautiful its campaign is.

I think about whether there’s something there.

A particular way of seeing the world. A relationship with people. A choice that feels specific enough that another company couldn’t simply put its logo on it.

Sometimes it’s an unusual sense of humour. Sometimes it’s the founder’s obsession with a detail most people wouldn’t notice. Sometimes it’s a community that recognises itself in the brand.

You can feel that someone made a decision.

Of course, this is subjective. What feels interesting to me might feel completely irrelevant to you. That’s part of culture, too.

But I think brands get into trouble when they try to remove that tension. When everything needs to be familiar enough to approve, broad enough to appeal to everyone, and easy enough to produce again.

AI fits very comfortably into that process.

It can also help us explore something unexpected. But we have to want to go there in the first place.

I understand why founders want it

If you’re scaling a beauty or wellness company, there’s a lot to make.

Product pages. Retail presentations. Campaigns. Emails. Social content. Then different versions of all of those things.

And if you’re a small team, the same people are probably carrying most of it.

So yes, I understand the appeal of a tool that helps you move faster. I don’t think every repetitive task needs to become a meaningful creative experience. Some things just need to get done.

There’s evidence that AI assistance can help. A study involving 5,179 customer-support agents found an average productivity increase of nearly 14%, with greater benefits for less experienced workers. That’s a particular workplace setting, so I wouldn’t apply the number to an art director’s workflow. But it’s a real example of useful assistance. NBER, Generative AI at Work

What interests me is what happens next.

If the team saves time, do they get to think more carefully? Talk to customers? Develop an idea they previously couldn’t afford to explore?

Or does the company immediately expect twice as much content?

Because those are very different outcomes for the people doing the work.

Better work can still become more similar

ai slop
ai slop
ai slop

There’s a study I find useful here because it complicates the conversation.

Researchers asked people to write short stories, with some participants receiving AI-generated ideas. Those ideas improved assessments of individual creativity, especially for writers who were less creative to begin with.

But the stories also became more similar to one another.

So the individual work improved while the overall range narrowed. Doshi and Hauser, Science Advances

This was a short-story experiment. It doesn’t prove that every AI-assisted brand will become generic.

Still, I think it gives founders something worth questioning.

If we’re all asking similar tools for similar things, using similar references, how different are we expecting the results to be?

Think about a hypothetical skincare campaign. Soft lighting. A floating bottle. Glossy liquid. A sentence about transformation.

It might look good. I might even like it.

But if I cover the logo, would I know whose campaign I’m looking at?

That’s where my interest shifts from the quality of the image to the strength of the brand.

Does nobody care, as long as it’s cheap?

I understand the frustration behind that question.

When you care about the work, watching someone accept a weaker version because it arrived faster can feel dismissive. Especially when you can see what’s wrong and know what it would take to fix it.

But I’d be careful about saying people don’t care.

A customer might not care which software made an image. They can still care whether the product looks accurate, whether the message feels manipulative, or whether the brand has anything to say.

And business leaders may be working with measures that make the saving visible immediately while the cost to the brand takes longer to appear.

You can count how many assets were delivered this month.

It’s harder to measure the moment someone stops finding your brand interesting.

That doesn’t mean the loss is imaginary. It means we need to pay attention to more than production volume.

Beauty and wellness make this more personal

These industries speak to people about their bodies, their confidence, and sometimes their health.

So I don’t think every generated image can be treated as decoration.

An imaginary landscape around a fragrance bottle is one thing. A convincing skin transformation placed beside a serum is another.

What is someone being asked to believe?

For brands marketing in the United States, the FTC’s health-products guidance explains that images can communicate implied claims, and those claims need support. An image looking convincing doesn’t establish that a product delivers the result. FTC, Health Products Compliance Guidance

The same concern applies to customer stories. A fictional person can’t truthfully describe an experience they never had. The FTC’s consumer-review rule addresses fake testimonials, including AI-generated ones. FTC, Consumer Reviews and Testimonials Rule Q&A

Dove has drawn a specific boundary: its Real Beauty Pledge includes a commitment never to use AI imagery in place of real women. I’m interested in that as a positioning decision. It makes a connection between what the brand says it believes and what it permits itself to produce. Dove, Real Beauty Pledge

Other companies may draw different boundaries.

But they should be able to explain them.

What needs to remain real for someone to trust your promise?

My taste doesn’t make me exempt

Personally, I can generate a great amount of work. My taste and skills help me shape it.

But I don’t want to assume that because I’m making the final decisions, the process can’t influence me.

If I spend enough time choosing between suggestions from the same systems, do I start getting used to what they offer?

Do certain compositions begin to feel natural simply because I keep seeing them?

I think that’s worth asking.

Cultural research requires me to look beyond the image. Who is participating? What does something mean to them? Where did it come from? What changes when a brand adopts it?

A visual reference can show me how a community presents itself. Understanding that community takes more attention.

So I want to keep encountering things outside the tools: people, places, art, materials, conversations.

That gives me something to bring into the work.

And I want to assess AI both subjectively and objectively.

Does this feel right to me? Is it accurate? How long did it actually take to finish? Do we have the rights to use it? How are people interpreting it?

My opinion matters. It also needs to be examined.

When someone feeds your work into AI

This part feels particularly uncomfortable.

You deliver something to a boss or client. They upload it to a tool and generate more work from it. Maybe the new versions lose what you were trying to do. Maybe they’re published without your involvement.

I understand why that can feel like your contribution has been reduced to raw material.

But before calling it theft, I’d want to establish what was agreed.

Who owns the work? What adaptations are permitted? Was uploading it allowed? Are there licensed images or other people’s rights involved? Who approves the new versions?

Using a file as a reference and training a model on it are also different uses. We need to be clear about which one is happening.

The legal question won’t always match the ethical one. My own position is that this deserves a conversation before delivery.

If you’re commissioning someone for their distinctive contribution, be clear about how you intend to extend it. Give them a chance to understand the relationship they’re entering.

That seems like a reasonable place to start.

What would I ask a company to do?

First, decide what you’re trying to improve.

If AI helps you organise information or explore an early concept, useful. If it creates more corrections than it saves, count those hours too.

Keep someone responsible for the claims, the representation, and the final work. And give that person the authority to say something isn’t ready.

Be clear about what can be uploaded and adapted.

Then look at where the saved time goes.

I’d want some of it to support better customer understanding, original creative development, and a team that has room to think.

Because if every efficiency becomes another content requirement, we should be honest about what we’re building.

I don’t know whether human-made work will automatically become more valuable as generated material becomes more common. I can imagine people seeking out evidence of care, authorship, and real experience. I can also imagine companies continuing to prioritise convenience.

Neither outcome is guaranteed.

What I do know is that generating more doesn’t resolve the question of why someone should care.