Why AI’s limitless creation triggered an authenticity reset

• Generative AI has made content creation faster and cheaper, leading to an oversaturation of similar marketing materials, which risks diminishing brand authenticity and consumer engagement.

• Research indicates a growing disconnect between ad executives' positive perceptions of AI-generated content and actual consumer sentiment, highlighting a backlash against the uniformity of AI outputs.

• Brands like He Gets Us and Porsche have opted for manual craftsmanship in their advertising to differentiate themselves in a landscape filled with AI-generated content, signaling a return to authenticity.

• Brands can use AI to stress-test creative concepts before production, allowing them to identify which ideas resonate and ensuring that marketing efforts are grounded in genuine consumer insights.

Why it matters

As generative AI dramatically lowers the cost and speed of producing marketing content, the risk for brands is not scarcity but sameness. Marketers need to rethink how AI is used in the creative process, shifting it from a production engine to a tool for testing ideas, discovering what truly resonates and scaling work that has already proven its value.

Key takeaways

• Imperfection, craft and storytelling are becoming markers of authenticity in an environment saturated with AI-generated content.

• Use AI upstream to test ideas, not just produce assets. AI can generate large numbers of concept variations quickly, helping creative teams diagnose whether an idea is strong before investing in production.

• Discover creative guardrails through experimentation. Stress-testing concepts across formats and contexts reveals which creative elements truly hold a brand idea together. These insights can become practical brand constraints.

• Testing concept variations with real or simulated audiences helps ensure creative work resonates before large media or production investments are made.

• Scale only after meaning is proven. Test concepts, identify what works, encode those learnings and only then scale production.

Earlier this year, the Super Bowl became an unlikely referendum on AI. Brands that had spent the past two years racing to automate their creative output made a different choice: He Gets Us shot their spot on film. A few weeks earlier Porsche ran hand-drawn illustrations for its holiday ad. And Panda Express commissioned human animators for the Lunar New Year. In an advertising environment saturated with AI- polished content, manual craft has become the differentiator.

This is not a niche cultural correction. According to research from the IAB and Sonata Insights, while 82% of ad executives believe Gen Z and millennials feel positively about AI-generated ads, only 45% of those consumers actually do. That gap has been widening.

The backlash reveals where production discipline broke down and what a smarter path forward looks like.

When the cost of creation drops to zero, the cost of attention rises

AI removed friction from making. Copy, images, video, campaign variants—all of it became fast, cheap, and scalable. The natural response was volume. If generating a thousand asset variations costs almost nothing, why not generate a thousand?

Here is the problem that volume thinking ignores: audiences do not evaluate content in isolation. They evaluate it against everything else competing for their attention at the same moment. When every brand has access to the same tools and produces content through the same processes, the outputs converge. The very smoothness that AI optimizes for becomes the dead giveaway that something was manufactured rather than made.

Brands are now actively requesting imperfections from creators like an unmade bed, dishes in the sink, hair that isn’t perfect. Not because messiness is inherently valuable, but because it signals that a human made a judgment call. Imperfection has become a proxy for intention. And in an environment of limitless creation, intention is scarce.

And consider the resurgence of the human storyteller. The job title “storyteller” doubled on LinkedIn between 2024 and 2025. Netflix posted a communications role at $775,000. Anthropic tripled its comms team. It’s a blowback from the proliferation of AI slop.

Abundance has made authenticity harder to demonstrate. Those are different problems, and they require different solutions.

Stop using AI to produce. Start using it to pressure-test.

The core mistake most brands have made with AI is deploying it downstream to manufacture finished content at scale rather than upstream, where it can do something more valuable: tell you whether an idea is worth making in the first place.

Before a single finished asset gets made, use AI to generate hundreds of variations of the concept. Push it into different contexts, formats, aesthetics, and cultural settings. The goal is diagnosis, not output. Does the idea hold when you stretch it? Or does it go vague the moment you move it off the original reference image?

Consider a sportswear brand testing a campaign concept built around “quiet confidence. ” The team generates hundreds of AI variations, like athletes in solitary training environments, crowded urban settings, high-pressure competition moments, everyday commutes. In some contexts, the concept lands with clarity. In others it reads as detachment, or worse, arrogance. That drift is the information. It tells the creative team exactly which environments and visual conditions carry the idea and which ones hollow it out. The concept either earns the next step or it goes back to development. Either way, the brand has learned something before spending a dollar on production.

This is what AI was built for in a creative context: generating the raw material that makes human judgment possible at a scale no human team could manage manually.

Use what survives to set the rules, not a style guide

Most brand guidelines describe what a brand looks like. They specify typefaces, color palettes, approved photography styles. What they rarely capture is why certain creative choices work and others don’t. that knowledge has traditionally lived in the heads of senior creatives who have worked on the brand long enough to develop instinct.

Stress-testing at scale changes that. When you run a concept through hundreds of variations and observe which ones hold, you are no longer guessing at the load- bearing elements. You are discovering them. The specific lighting temperature that makes the brand feel warm rather than clinical. The compositional rule that creates tension without chaos. The tonal register that signals expertise without condescension.

Those discovered constraints become your actual creative guardrails, earned through evidence rather than asserted through consensus. A designer is working from them is working from proof, not from someone’s strongly held opinion about what the brand should feel like.

The practical output here is a set of non-negotiables that are specific enough to be actionable. Not “our brand feels warm. ” But: “our brand requires golden-hour color temperature, centered composition, and at least 30% negative space, and we know this because everything else we tested fell apart. ”

Validate against the market before you commit to it

Internal stress-testing tells you which ideas survive creative pressure. It does not tell you which ones resonate with actual people. This requires a brand to test the concepts that survived internal review against real audience response before committing production budget.

The same AI tools that generated your variations can help construct synthetic audience models built on behavioral data, or proxies that simulate how different segments respond to the same concept across different formats. Run the survivors from your stress test through those models. Which versions retain their meaning when the audience changes? Which ones depend on assumptions that only hold for one segment? The answers narrow the field further, and they do it before a media buy has been made.

For teams without access to synthetic modeling, a lightweight version of this exists in paid social. Put three or four of your strongest concept variations into a small dark post test, with real creative, real audiences, minimal spend. What you are measuring is not click-through rate. You are measuring whether the intended meaning is landing. Comments, saves, and shares tell you something about resonance that impressions never will.

The point is to close the gap between what the creative team believes works and what the market actually responds to before scale makes that gap expensive.

Encode what you’ve learned into the system

Once you know which elements are load-bearing, you can train your AI tools on those specifically based on the subset of work that survived your stress test.

The result is an AI that produces on-brand work by default because it has been trained on the evidence of what your brand actually is. A designer prompting it for a new product image does not need to specify the lighting, the composition, or the tonal temperature. The model has internalized those constraints. It starts from your brand’s creative logic rather than from a statistical average of everything it has ever seen.

This is the difference between using AI to replicate a brand and using AI to understand one. Replication produces volume, but understanding produces work that holds.

For teams managing multiple sub-brands, agency relationships, or high-velocity content calendars, the encoded model becomes a shared creative standard that travels with every brief. This enforces consistency through the outputs themselves, not through review cycles and revision rounds

This is the difference between using AI to replicate a brand and using AI to understand one. Replication produces volume, but understanding produces work that holds. For teams managing multiple sub-brands, agency relationships, or high-velocity content calendars, the encoded model becomes a shared creative standard that travels with every brief. This enforces consistency through the outputs themselves, not through review cycles and revision rounds.

Scale is the reward, not the starting point

The authenticity reset is a call to use AI in the right sequence. Test first. Discover what holds. Encode what you learn. Validate against the market. Then scale with confidence that what you are scaling has already been proven worth making.

Authenticity did not disappear when AI arrived. It became harder to demonstrate. Limitless creation is now the baseline. Brands that stand apart design for intention before they designed for scale.

Featured in WARC.