AI-generated advertisements outperformed ads created by human designers in a live campaign, and the advantage still held 18 months after the initial run.

That is the headline result from a study reported by Tech Xplore, and it lands directly in the middle of an argument the advertising industry has been having since generative tools went mainstream: can machines replace — or at least match — professional creative teams on real commercial work, not just lab benchmarks?

What did the campaign actually show?

The study pitted AI-generated ad creative against ads produced by human designers in a live campaign setting. This matters. Plenty of experiments compare machine and human output in controlled conditions, where panels of raters score images in isolation. A live campaign is different: real audiences, real placements, real performance data.

According to the reported findings, the AI-generated ads came out ahead. The researchers then went back and checked whether that advantage lasted. It did. Eighteen months later, the machine-made ads still outperformed the human-designed ones.

The exact performance metrics — click-through rates, conversion figures, spend levels or campaign budget — were not specified in the report as surfaced here. Mart Signal readers should treat the direction of the result as confirmed and the magnitude as unquantified until the underlying paper is reviewed directly.

Why does the 18-month follow-up matter?

Most coverage of generative AI in advertising stops at the first result. The durability question is the harder one.

Creative effects in advertising often decay. An ad that wins on novelty in month one can fatigue by month six. If AI-generated creative had simply exploited a temporary novelty effect — audiences clicking because the output looked unfamiliar — the advantage should have eroded over time.

It did not. The gap was still measurable a year and a half later, which suggests the performance difference was not a short-lived artifact but something more stable in how the AI-produced ads communicated or converted.

Who does this affect?

The results speak to three groups directly:

• In-house creative teams at brands that currently brief human designers for performance campaigns
• Agencies selling design hours against increasingly capable generative tools
• Ad-tech vendors building automated creative pipelines into buying platforms

For operators running performance campaigns, the practical question is no longer whether AI creative can compete. On this evidence, it can win. The question is cost, workflow and control.

The skeptical read

One live campaign is one live campaign. The study result is notable because it is field evidence rather than a panel test, but a single campaign cannot tell you:

• Which product category, audience or platform was tested
• What skill level the human designers had
• What tool produced the AI creative and at what cost
• Whether the result replicates across markets

None of those details appear in the surfaced report. Without them, the finding is a strong signal, not a playbook.

Still, the direction of travel is consistent with what the market already shows. Major ad platforms have pushed automated creative features for years, and advertisers have adopted them at scale because the numbers hold up in dashboards, not because of vendor marketing.

What should operators take from it?

Test AI creative against your human output in your own campaigns, with your own performance data as the judge. Eighteen months of sustained advantage in a live setting is a result worth replicating — or disproving — on your own spend before anyone restructures a design team around it.