AI in advertising: Synthetic models on the rise

“What image of beauty should AI learn from us?” This is the central question of the current Dove campaign. In the film called “The Code,” a woman sits at the computer and asks the AI ​​software to show her the “most beautiful woman in the world.” Then she asks the question again with the addition “like in a Dove ad.” Suddenly the images are more diverse and appear more authentic.

For 20 years, the cosmetics manufacturer has positioned itself as the brand that promotes “true beauty”. In the beginning, Dove pioneered diversity in advertising and consciously selected models who did not meet the advertising industry’s beauty standards.

The question posed by Dove is extremely relevant. Image-generating AI software is now so efficient that its products are practically no longer recognized as synthetic, the study has found “AI Humans in Brand Communication 2024” from the Stuttgart Media University in collaboration with the WongDoody agency.

While Dove is committed to “never using AI to create or distort images of women,” other brands have started replacing human models with AI-generated ones. This summer, the Spanish clothing manufacturer Mango launched its first AI-generated campaign with media impact. A young woman presents the brand’s clothes. It is not clear that the woman was created digitally.

AI-generated models are also used in Switzerland. Calida presented the first synthetic image series in June. The aim of the project is to “explore the limits and possibilities of artificial intelligence in the visual representation of laundry.” In a press release, the product manager, Janine Weiz-Bühler, was “enthusiastic”. AI makes it possible to supplement image production “spontaneously and specifically,” wrote the underwear manufacturer.

The St. Gallen design agency Kreuzer Design is also increasingly using AI. «We can create consistent image series, but also control details according to customer requirements. For example, we can place people freely or pose them. Replacing people in the picture, aging them, changing their ethnicity, diversity, etc. Detailed shots are no problem either,” explains owner Alexander Kreuzer to persoenlich.com.

From his point of view, the new technical possibilities will completely change the industry. «A shoot for an internationally operating client usually involves extreme effort: models of different ethnicities have to be ordered, model rights have to be renewed, the locations have to be approved, the photographer and stylists have to be paid. All of these things are now gone.”

Apart from the organizational questions and the new business models that will emerge, ethical questions also arise. First, should AI-generated images be declared as such? According to one Opinion poll According to the market research platform Appinio in August, 81 percent of Germans demand that this content be clearly marked.

“Labeling is easy as long as the use of AI tools is clear,” notes Johan Rochel, co-founder of Ethix, the laboratory for innovation ethics in Zurich, when asked by persoenlich.com. “But AI tools are increasingly being integrated into software, so that you are no longer aware that you are working with AI.”

Risk for the brands

For the ethics expert Rochel, the crucial question is: “Do you want to intentionally manipulate the audience, like deepfakes do?” In his opinion, there is a risk for brands that the use of AI images will damage their credibility or raise doubts among consumers. It’s about a tactical balance between cost reduction and the image you want to project.

There is also concern that AI-generated images would portray people as perfect and standardized, according to the AI-Human study from Stuttgart. This concern is not new. Dove wanted to counteract this trend with the claim of “true beauty”.

What’s new is that AI allows the images to be created exactly the way the brand wants, notes Johan Rochel. There are no longer any external factors. The brands cannot absolve themselves of responsibility.

Industry code “unnecessary”

So far acceptance seems to be high. In the Appinio survey, 62 percent of respondents understand the use of AI in the Mango campaign for cost reasons. Fiction and illusion are part of the advertising industry, explains Johan Rochel. This allows a certain amount of freedom with the use of AI.

An industry code of conduct could make sense, says the ethics expert. Many media outlets have introduced guidelines in journalism. In advertising, the five center-left parties signed an AI code last year. They avoid using AI-generated content for negative campaigns. The idea arose after a controversy surrounding an FDP poster that showed an AI image of climate activists. The advertising client association SWA currently believes that an industry code is unnecessary. “Common sense is enough, because there are already more than enough laws and rules for advertising,” writes SWA director Roland Ehrler when asked.