AI Avatars Still Can’t Beat a Real Person’s Trust — What Myntra’s and Others’ Virtual Influencer Experiments Reveal

Somewhere in a Mumbai studio last year, a team of designers put the finishing touches on a face that did not exist. Every eyelash was rendered, every strand of hair simulated, every micro-expression calibrated against thousands of hours of human video. The result was uncanny, polished, tireless and, on paper, exactly what a brand marketer dreams of: a spokesperson who never ages, never has a bad day, never asks for a higher fee and never says the wrong thing at the wrong moment. And yet, when the campaign numbers came in, the uncomfortable truth surfaced again, the one the industry keeps rediscovering every time it falls in love with a new avatar. People buy from people. Even when those people are famous strangers on a screen, they still register, somewhere in the reptilian core of the brain, as real. A synthetic face, however exquisite, does not.

This is the paradox sitting at the centre of India’s latest experiment with virtual influencers, and Myntra’s foray into AI-generated brand ambassadors has become something of a case study in both the promise and the limits of the format. The fashion e-commerce giant is hardly alone. Globally, brands from Prada to KFC have dabbled with digital humans and AI-fronted campaigns, chasing the same cocktail of novelty, cost efficiency and creative control. But the pattern that keeps emerging, across geographies and categories, is strikingly consistent. Virtual influencers generate attention. They rarely generate trust. And in a category like fashion and lifestyle e-commerce, where purchase decisions live or die on perceived authenticity, that distinction is not academic. It is the whole ballgame.

The economics were always going to be seductive

Consider the arithmetic from a brand’s vantage point. A top-tier Bollywood ambassador can command anywhere between two and ten crore rupees for an annual retainer, before production costs, before the inevitable scheduling battles, before the risk that a single controversial tweet or a poorly timed personal scandal unravels months of campaign equity overnight. A virtual influencer, once built, can be deployed across a hundred creative variations simultaneously, dressed in every outfit from a new drop within hours, speaking in a dozen regional languages without a dubbing artist in sight, and never once threatens to embarrass the brand at a press event. For a category like fast fashion, where the content calendar is relentless and the margin for error on brand safety is thin, that pitch is genuinely compelling on a spreadsheet.

It is also, several marketers now privately admit, a pitch that undersells how much of a fashion purchase decision is emotional rather than logical. When a shopper sees a real influencer wearing a kurta or a pair of sneakers, there is an implicit transaction happening beneath the surface. The viewer is not just evaluating the garment; they are evaluating the wearer’s life, their choices, their perceived taste, and asking, consciously or not, whether that life is one they would like a sliver of. A virtual avatar can simulate the aesthetics of that life with remarkable fidelity, but it cannot simulate the stakes. There is no risk in an AI model’s endorsement, and audiences, it turns out, are far more perceptive about risk than brands often give them credit for. If the ambassador has nothing to lose by recommending a product, the recommendation itself starts to feel weightless.

Myntra’s own experimentation reflects an awareness of this tension rather than a naive dive into it. Rather than positioning its AI avatars as replacements for human influencers, the brand has largely used them in supporting or novelty roles, style demonstrations, quick-turnaround seasonal content, experiential features that let shoppers visualise outfits, functions where the technology’s strengths in speed and scale are genuine assets rather than substitutes for credibility. That distinction matters enormously, and it is one that a growing number of Indian marketers are beginning to internalise. The winning use case for AI avatars is not “please like and trust this being.” It is “please use this tool to see the product more clearly, faster, in more contexts, at lower cost.” Those are fundamentally different jobs, and conflating them is where most virtual influencer campaigns quietly stumble.

What the global scoreboard actually shows

Look beyond India and the evidence sharpens further. Lil Miquela, arguably the most famous virtual influencer on the planet, has amassed millions of followers and lucrative partnerships with luxury houses over nearly a decade of existence, an achievement that speaks to genuine creative craft and sustained novelty value. But dig into engagement quality rather than follower counts, and a gap opens up. Comment sections under her posts skew heavily toward commentary on the technology itself, marvelling at the render, questioning the ethics, debating whether she counts as “real” content, rather than the kind of aspirational or advice-seeking engagement that drives actual purchase intent among human creator audiences. The conversation stays about the avatar. It rarely graduates to being about the product.

That is a subtle but critical failure mode. Human influencer marketing works, at its best, because the creator becomes invisible to the transaction; the audience stops thinking about the creator as a creator and starts imagining themselves in the product. Virtual influencer content, by contrast, tends to keep the spotlight fixed on its own artifice. Every post is implicitly also a demo of the technology, and demos, however impressive, rarely convert as effectively as an anecdote from a trusted peer. KFC’s brief AI-generated Colonel Sanders campaign in the West drew headlines and think-pieces in equal measure, but the discourse it generated was almost entirely about the novelty and the discomfort of the uncanny valley, not about fried chicken. Attention and affinity, it turns out, are not the same currency, and brands that mistake one for the other end up with impressive reach numbers sitting atop disappointing conversion.

There is a deeper structural reason for this, rooted in what trust actually is. Trust, in the marketing sense, is essentially a bet that someone else has skin in the game on your behalf, that a recommendation reflects genuine preference rather than a script. Human influencers, whatever their commercial arrangements, still carry biography, reputation and continuity that audiences can cross-reference. A fitness creator who has documented a genuine transformation over years lends credibility to a supplement endorsement precisely because the audience has watched the story unfold in real time, with visible stakes. An AI avatar has no biography that predates the brand deal. It has no continuity that was not manufactured for commercial purpose. Audiences may not articulate this gap in technical terms, but they clearly feel it, and the data on engagement composition bears that instinct out repeatedly.

Where the technology genuinely earns its keep

None of this means the virtual influencer experiment has been a wash, and it would be a mistake for marketers to read the trust gap as a verdict against the technology altogether. The more instructive lesson is about placement within the funnel rather than abandonment of the tool. AI avatars and digital humans have proven their worth in functions that do not require the audience to suspend disbelief about sincerity: rapid-turnaround product visualisation, localisation of a single creative asset into multiple languages and cultural contexts without reshoots, always-on customer service personas, virtual try-on experiences where the value proposition is explicitly technical rather than emotional. In each of these cases, the audience is not being asked to trust a person. They are being asked to use a tool, and tools are judged on utility, not on sincerity.

This reframing has quiet but significant implications for how Indian brands should be budgeting for the technology going forward. Rather than pitting AI avatars against human ambassadors in a zero-sum contest for the hero campaign slot, the more sophisticated media strategy treats them as complementary layers of the same funnel. A human face anchors the brand story at the top of the funnel, where aspiration and trust do the heavy lifting. AI-generated content handles the high-volume, high-frequency middle and lower funnel work, product demonstrations, size and fit visualisation, festive collection previews, regional adaptations, where speed and scale matter more than emotional resonance. Myntra’s approach, in this light, looks less like an early misstep and more like a template other category leaders are likely to converge toward, whether or not they credit the fashion platform for getting there first.

There is also a generational dimension worth watching closely. Gen Z audiences, who grew up alongside filters, deepfakes and CGI influencers as a baseline feature of the internet rather than a novelty, show somewhat more tolerance for synthetic personas than older cohorts, provided the artificiality is transparent rather than concealed. The moment a brand tries to pass off an AI avatar as human, or obscures the distinction to manufacture false intimacy, the backlash tends to be swift and disproportionate, precisely because the perceived deception violates the implicit contract of authenticity that underpins all influencer marketing. Transparency, counterintuitively, may be the feature that eventually lets virtual influencers earn a measure of trust human ambassadors take for granted. An avatar that openly says “I am an AI-generated character built to show you this collection” is playing an honest game. One that hides its nature is not, and audiences increasingly know the difference.

For agencies and brand teams mapping out the next eighteen months of creator strategy, the practical takeaway is less about picking a winner between human and synthetic and more about precision in assigning each format to the job it can actually do. Trust is not a technology problem that better rendering will eventually solve; it is a relationship problem, built on perceived stakes, continuity and honesty, none of which improve simply because the pixels get sharper. Myntra’s experiments, and the broader global scoreboard of virtual influencer campaigns, point toward a mature middle path rather than a binary verdict. AI avatars are not going to replace the real person your audience wants to believe in. But used well, in the roles built for utility rather than intimacy, they may finally justify the hype that has surrounded them since the first uncanny face blinked to life on a marketing team’s screen.