Synthetic Media Authenticity Gradient
Tolerance for AI-generated content varies sharply by audience: the 20-to-45 age cohort questions authenticity and is put off by synthetic media, while minors and older audiences mostly no longer care. Crossed with a product axis, commodity goods tolerate synthetic content while premium goods demand pixel-perfect realism.
The finding
Alex Mashrabov reports this as an internal data finding from his own company rather than a general opinion, and it runs along two axes that cross each other. On the age axis, the pattern is counterintuitive: the twenty-to-forty-five cohort "is a little bit concerned about authenticity and they like to question authenticity," in Mashrabov's words, the most skeptical group, while minors and older audiences "typically don't care anymore." The implication cuts against a common assumption that synthetic media is primarily a younger-generation play, since the data instead suggests it reaches well beyond that group while specifically struggling with the prime spending cohort most inclined to interrogate it, meaning a simple pairing of synthetic content with younger audiences is probably not exactly right.1
On the product axis, commodity goods tolerate obviously synthetic, slightly imperfect content well, the familiar waterproof-headphones-on-TikTok aesthetic used for everyday consumer products, while premium and aspirational goods demand pixel-perfect realism, because what they are actually selling is not just a product's function but a specific feeling and aspiration that a visible flaw can undermine entirely. That top rung of the product axis overlaps with the hardest end of what generative models can currently render convincingly.1
Put together, these two axes describe a served-market map: synthetic media performs best where commodity goods meet audiences that do not interrogate authenticity closely, and performs worst where premium or aspirational goods meet the skeptical twenty-to-forty-five cohort.
Why it matters
This functions as the audience-acceptance complement to a separate technical difficulty hierarchy for generative models: one axis bounds what a model can actually render, and this one bounds what an audience will actually accept, and a complete go-to-market read on any synthetic-media product needs both together. It also corrects a common targeting assumption, that synthetic content is inherently a younger-generation-native play, with real internal data, redirecting synthetic-media budgets toward where acceptance is genuinely highest rather than where it is merely assumed to be highest.1 It also bounds where AI-generated brand ambassadors are likely to work well, thriving in the tolerant zones of the map and provoking backlash in the skeptical or premium zones, and it helps explain why authentic affiliate marketing and taste-driven, human-fronted marketing keep their value specifically inside the skeptical, high-spending twenty-to-forty-five cohort, exactly where the gradient runs steepest.
Tensions
The age-cohort findings and the claim that older and younger audiences no longer care are internal data relayed in a single interview, directionally useful but not independently, externally validated, and cohort attitudes are also likely to shift quickly as continued exposure normalizes synthetic media over time. The gradient itself may flatten in either direction: today's skeptical twenty-to-forty-five cohort could grow more tolerant simply through repeated exposure, or could grow more skeptical instead as detection and labeling tools improve, and which direction wins out is not yet knowable from the data described here.
Practiced by
Connections
Loading connections…
References
- 01
Alex Mashrabov on Higgsfield: Generative Video & Consumer AI (Venture with Grace)
Alex Mashrabov · interview · 2026
Related