Is there a model worth killing human creativity for?

The question every agency is asking right now is whether AI can make the ad. It can. That was never the interesting question — a tool that generates a competent thirty-second spot from a prompt has existed for a while now, and the demos are genuinely good. The question that pays the mortgage is the next one: does the ad still do its job? And on that one, the effectiveness data has an answer most of the industry doesn’t want to hear.

And to be clear about who’s talking: I’ve spent a career in this business, and I run an AI company now — we use these generative tools every day, on real work. So this isn’t a creative defending his turf from a technology he’s never touched. It’s an operator who runs the machine telling you, plainly, what the machine does to the one number that actually matters.

What the machine actually changes

Kantar put AI-generated ads in front of people and read their faces. The finding is precise and it is awkward: viewers react more to AI-generated advertising, but net positivity comes out lower. They feel something — often the wrong thing. The uncanny valley doesn’t announce itself; it just registers as a faint, unplaceable discomfort that drags the response down. Where the generative work was genuinely seamless, more than 40% hit top-tier branded cut-through. Where it was obviously machine-made, it performed worse than the human benchmark.

So the variable was never whether you used AI. It was how — and “how” is precisely the part the cost-cutting case is trying to skip. The pitch is “same creative, a tenth of the price.” The data says you got different creative, and the difference is sitting in the column that decides whether it sells.

Emotion is the multiplier, and it’s measurable

This is where the romance about “the human spark” turns into arithmetic. System1, working with the IPA’s effectiveness data, found that adding a measure of creative quality — their Star Ratings — to media spend nearly doubles the accuracy of predicting which brands grow share. Mark Ritson, reading the fused System1–Effies dataset, put the top end of it plainly: running distinctive, emotional, consistent advertising can drive something like a 12× profit multiplier over the alternative.

Read that again, because it’s the whole argument. The emotional register isn’t the garnish on an effective ad. It is the engine. And it’s the first thing to thin out when you hand the work to a system that can simulate the surface of feeling but has never had one.

AI didn’t make the ad cheaper. It made the expensive part — the feeling — optional. That was the part paying you.

The trust tax nobody put in the model

Then there’s the audience’s side of the ledger, and it’s moving fast. The share of consumers who say heavy AI use would decrease their trust in a favourite brand doubled in a year, from 20% to 40%. Among Gen Z it’s 54%. Ninety percent want AI use disclosed — and disclosure itself measurably lowers how people feel about the ad. That’s a trap with no clean exit: hide it and you’re exposed, declare it and you take the hit. McDonald’s and Coca-Cola have both already eaten public backlash for AI campaigns that read as cheap to the people they were aimed at.

The authenticity premium isn’t a soft value. It has a price tag, it’s rising, and it lands straight on brand equity — the one asset on the balance sheet the CFO can’t rebuild in a quarter.

“But it scales”

Here’s the strongest counter, and it’s a real one: a human creative team doesn’t scale, and an AI pipeline scales infinitely. True. But scale is a multiplier, not a virtue. Multiply work that doesn’t move anyone and you don’t get growth — you get more impressions of the thing that wasn’t working, delivered faster and cheaper than ever. Reach times zero is still zero. Infinite distribution of forgettable is just a more efficient way to be forgettable.

Build human × agent — not human-instead-of

None of this is an argument against AI in the creative process. I’m building an agentic AI company; I want agents in the studio. It’s an argument about where you point them. There’s a boring 85% of advertising production — versioning, resizing, trafficking, localization, the endless re-keying tax — that is begging to be handed to agents, and the people doing it by hand right now will be thrilled to let go. Then there’s the 15% that carries the feeling: the idea, the tension, the line that lands. That stays human, and you protect it with everything.

The teams that win the next decade won’t be the ones who fired the creative department to afford the tooling. They’ll be the ones who armed it — human craft on the part that moves people, agentic throughput on the part that never did. That pairing is the real edge, and the talent to build it isn’t concentrated in two zip codes on one coast. It’s everywhere the industry keeps underestimating, including here.

So, is there a model worth killing human creativity for? No — and not for sentimental reasons. Because the part the machine can’t make is the part that works. Arm the humans. Don’t replace them. The day you optimize the feeling out of the ad is the day you optimize the growth out with it.