I Followed Someone Who Doesn't Exist

A while back, while scrolling through social media, I came across an influencer called Lil Miquela.

Gorgeous, great fashion sense, even her own music career. More than three million followers.

Then one line stopped me cold: this person doesn't exist.

She's a virtual influencer — made entirely with AI. Even that signature little freckle of hers was designed too.

Wow.

But the bigger surprise came later. In 2018, the French luxury house Balmain dropped the Kardashian family — their partners for years — and replaced them with three virtual models: Shudu, Margot, and Xhi. During the pandemic, the World Health Organization also brought in a virtual influencer named Knox Frost for a public-health campaign. This "person" has over a million followers on Instagram. And KFC, LVMH, Nike, Netflix, and Samsung have all shaken hands with virtual influencers too.

How does someone who doesn't exist wield all that power?

What Is a Virtual Influencer?

Put simply: a "person" made with AI who lives in your following list.

Take Shudu, who calls herself "the world's first digital supermodel" and has worked with major magazines and athletic brands. Her path is different from Lil Miquela's: Shudu goes for pure visual delight, like a fashion editorial that can talk; Lil Miquela is carefully packaged as a "real person" in her twenties — music career, worries, opinions — the kind of person you want to become.

Why do brands love them? Think about it: they're cheap, they're always available, no scandals, no aging, and they'll never blow up their image with one badly worded sentence.

Sounds like the perfect employee.

But here's the question: do consumers actually buy it?

Someone Did the Math

In 2022, a research team ran two experiments with several hundred US users, specifically comparing virtual influencers and human influencers: who influences consumers more?

The results were fascinating.

The first cut landed on trust. Virtual influencers scored 2.59 on trust; humans scored 3.28 (on a 7-point scale). Someone who doesn't exist starts out with a distance built in.

That distance has a name: "social-psychological distance." What does that mean? It's how close you feel the other person is to you, in your own mind. Human influencers post their daily lives, reply in the comments, interact with you — so they feel like a neighbor. A virtual influencer can look ever so real, but you know it's a piece of code. And with distance, trust thins out. The data confirmed it: it was exactly this distance that dragged trust down.

But the second cut revealed a twist.

Who do you want to follow? No difference. Virtual influencers didn't make people any less willing to follow them. Who do you think can recommend products tailored to you? No difference there either.

Why?

Think about it — you're long used to Netflix recommending movies, used to shopping platforms guessing what you'll like. People's trust in AI recommendations has already "spilled over" onto AI influencers. Trust built on real people and real stories has been picked up by AI.

The third cut was the most unexpected: virtual influencers were actually better at sparking word of mouth. After viewing their content, people were more willing to talk about it with friends, more willing to share. Probably because of the novelty. So you see, the WHO's move to use a virtual influencer for its public-health campaign was actually the right call — when it comes to spreading a message, they outperform even humans.

But There's One Condition — Don't Ignore It

The researchers also measured something else: does this influencer call their own shots, or are they tightly managed by their team?

The conclusion was blunt: human or AI, as long as an influencer is "controlled by someone," trust, follows, and word of mouth all slide downward.

Interestingly, virtual influencers are less susceptible to this penalty — expectations for code were low to begin with, so they don't take much of an extra hit for "being owned by a company." But once a human is exposed as a puppet on strings, they lose real fans.

There's also a group of people who fall especially hard for virtual influencers: those who crave being different. The more someone wants to live a one-of-a-kind life, the more easily they're drawn to a niche, virtual influencer. So if you're running a streetwear label, a niche subculture, or a personalization-driven product, virtual influencers could be a strong play.

But note: for these uniqueness-seekers, trust scores did not rise. For products that sell on trust, you still need a human.

So, Don't Rush to Replace Everyone

A word of caution for my fellow marketers: this research says "virtual influencers can hold their own," not "humans should be laid off."

The sensible playbook is roughly this: for spreading messages, chasing trends, and targeting niche audiences, virtual influencers are a good fit — content can be generated on the fly with any trend, and there's no persona risk. Push it to the extreme, and you could even build a dedicated virtual influencer for every single consumer — like virtually trying on clothes — showing each of them "a better version of yourself."

But for products that need trust behind them, a human is still your safe bet.

Back to the opening question: someone who doesn't exist — how do they have millions of followers?

For one simple reason: what influences you was never "whether they exist" — it's whether you're willing to treat them as a "person" too.