Every brand is using AI to make creative now. Scripts, images, voiceovers, whole video ads — generated in minutes.
And the ROAS is still flat.
The tools are not the problem. The misunderstanding of what creative actually does — that is the problem.
AI makes it faster to produce creative. It does not make it easier to produce creative that works.
Here is what is actually happening — and why most AI-generated ads are failing even the brands who think they are ahead.
01 AI is trained on what already exists
This is the thing nobody talks about when they're excited about AI creative tools.
These models learn from existing content. They learn from what has been made before. Which means, by definition, they are optimised to produce more of what already exists.
But here is the problem: what already exists is mostly average.
The feed is full of bland hooks, predictable structures, generic visuals. When you ask an AI to generate an ad, you are often asking it to produce a statistically probable version of every average ad it has ever seen.
That is not a creative brief. That is a mediocrity machine.
02 The feed rewards pattern interruption. AI produces patterns.
Ads work by breaking something. Attention. Expectation. The scroll.
The moment someone stops because they did not expect what they just saw — that is when an ad has a chance to do its job.
AI tools, as most people use them, do the opposite. They optimise for plausibility. For content that looks correct. For videos that seem like ads.
And content that looks like an ad is the easiest thing in the world to scroll past.
If your creative blends in, the algorithm cannot save you. Neither can your budget.
The brands winning right now are not making content that looks like ads. They are making content that earns a stop. That is a strategic decision, not a production decision. AI cannot make it for you.
03 Volume without judgment is just noise
The pitch for AI creative is usually speed and volume. More ads, faster, cheaper.
This sounds good. In practice, it just means more bad ads.
I have seen accounts with thirty active creatives, all AI-generated, all performing below benchmark. The problem is not the number of ads. It is that every single one is making the same mistake at scale.
More output does not fix a strategic problem. It amplifies it.
If you do not know what makes your customer stop, click, and buy — generating fifty variations of the wrong idea is not a strategy. It is noise.
04 The brands winning with AI are using it differently
There are brands seeing real results from AI in their creative workflow. But they are not using it the way most people think.
They are not asking AI to come up with the idea. They are using it to:
- produce variations of a concept that already works
- speed up production once the hook and angle are proven
- test formats without blowing the budget
- adapt winning creative for different placements and audiences
The thinking still happens before the tool gets opened. The strategy, the hook, the angle — that is human work. AI handles the execution.
That is a very different use case to "generate me a product ad."
05 What AI cannot replace
Good creative comes from understanding a specific person, in a specific moment, with a specific problem.
It comes from knowing why someone would actually stop. What language they use when they talk about their problem. What they have already tried. What they are afraid of. What they want to believe is possible.
That knowledge does not live in a prompt. It lives in customer reviews, support conversations, sales calls, and time spent genuinely understanding who you are talking to.
AI cannot do that research for you. And without it, the creative will always feel slightly off — technically competent, emotionally empty.
The best AI-assisted creative starts with a human who already knows what to say. The tool just helps them say it faster.
Final thought
AI is not going to ruin advertising. The brands who use it without thinking are going to ruin their own results — and they already are.
The tool is not the strategy. The volume is not the strategy. The strategy is understanding what makes someone stop, care, and act — and then building creative that does that.
AI can help you do that faster once you know what you are doing.
It cannot figure it out for you.
The brands who understand that will use AI to pull further ahead. Everyone else will just have more bad ads.