Photo-Illustration: Intelligencer

For a preview of how AI will collide with creative industries, look to advertising. Amazon, Google, and Meta have all started encouraging advertisers to use AI tools to generate ad copy and imagery, promising high performance, lower costs, and super-specific targeting. Now, brands are paying to advertise with AI-generated virtual influencers — synthetic characters that can offer at least some promotional juice at a fraction of the cost. According to the FT, influencers are concerned:

Aitana is a “virtual influencer” created using artificial intelligence tools, one of the hundreds of digital avatars that have broken into the growing $21bn content creator economy.

Their emergence has led to worry from human influencers their income is being cannibalised and under threat from digital rivals … But those behind the hyper-realistic AI creations argue they are merely disrupting an overinflated market.

“We were taken aback by the skyrocketing rates influencers charge nowadays. That got us thinking, ‘What if we just create our own influencer?’” said Diana Núñez, co-founder of the Barcelona-based agency The Clueless, which created Aitana. “The rest is history. We unintentionally created a monster. A beautiful one, though.”

The marriage makes sense. In a purely descriptive, nonjudgmental sense, AI produces inauthentic content; in a purely descriptive, nonjudgmental sense, the advertising industry is premised on inauthenticity. Advertising takes all the interesting, thorny, unsettling issues raised by generative AI and folds them into the same simple, eminently answerable question it always asks: Does this stuff sell products?

According to a case study by Meta, the answer is yes, sort of. An H&M ad campaign featuring Kuki, a virtual influencer and “always-on AI” chatbot created by a company called ICONIQ AI, resulted in an “11x increase in ad recall” — people who remember seeing an ad for a particular brand — over “ads with campaign video only,” resulting in a “91% decrease in cost per person recalling seeing ads.” Kuki is a cartoonish 3-D character; other virtual influencers, including the aforementioned and heavily sexualized Aitana, appear as photorealistic generated images, which could easily be confused, in the context of an ad, for real people.

It’s not quite right to say that virtual influencers are coming for the entire strange and maligned job of influencing. The advertising industry has been pushing the idea of synthetic talent for years, most notably in the form of Lil Miquela, a.k.a. Miquela, a.k.a. Miquela Sousa, a character who has appeared in ads for a range of prominent brands but whose influence as a character is both limited and inextricable from the metanovelty of the whole enterprise. Meta’s case study doesn’t describe a virtual influencer promoting products on its own account, to its own audience — the true core of the influencer business — but rather appearing in ads run by a company under its own accounts and promotional channels, wearing virtual clothes, as something between a spokesmodel and a mascot. The influencer version of grunt work, in other words. (Kuki’s own Instagram account now has a minuscule following and virtually no engagement.)

This makes for an odd new sort of advertisement that you might be likely to “recall”  for no other reason than not having seen such a thing before. It also situates virtual influencers less as a replacement for human influencers than as a new trick for juicing certain campaign metrics. This is the way Amazon suggests that AI-generated product photos get people to click ads at higher rates, even if those ads depict products in situations that defy physical reality.

This is no small thing, though — online advertising runs on such metrics, which measure influence in the bluntest possible terms. If featuring a cheap synthetic spokesperson in an ad produces measurably better results, brands will absolutely do so, until they either become the norm or the novelty wears off. To the extent this is a threat to human influencers, it’s a threat from the low end, replacing secondary sources of income in cases where an unfamiliar synthetic face is a sufficient replacement for an unfamiliar real face — cases where influencers are effectively working as internet-coded models, which, again, is a small but sometimes lucrative and easy part of the job, compared to cultivating an audience in the first place, or comprehensively influencing your fans to buy or do something. It’s also a useful if obvious way to think about how AI might encroach on a wide range of creative industries: identifying their most basic economic functions (converting clicks; scraping up attention; filling space; providing distraction) and attempting to peel them away for automation.