.png)
New research from NYU shows AI-created ads drive 19% higher CTR than human creative, but only when given creative freedom. Here's what that means for your creative process.
Groundbreaking research from NYU Stern and Emory has quantified what many performance marketers have suspected: generative AI creates more effective ad creative than humans. The study found that AI-created ads drove 19% higher click-through rates than campaigns featuring human-designed ads. When AI was also allowed to reimagine product packaging, the combined effect showed even stronger performance improvements.
This isn't just about AI being "good enough" anymore. It's about AI consistently outperforming traditional creative processes when deployed strategically.
But here's what makes this research truly revolutionary for enterprise marketers: the key to AI's superior performance isn't iteration and optimization, it's creative freedom.
The research reveals that visual AI thrives when given the freedom to generate cohesive, original concepts from scratch rather than working within the constraints of existing creative assets. This finding challenges the conventional wisdom that AI should be deployed as an enhancement tool for human creativity.
Instead, the data shows AI performs best as a primary creative engine, unencumbered by the need to preserve existing design elements.
The study tracked advertising performance across 105,999 impressions, comparing three distinct approaches: 1) human-designed ads, 2) AI-modified versions of those same ads, and 3) AI-created ads generated entirely from scratch.
When AI modifies existing human-designed ads (working within the constraints of someone else's layout, composition, and concept):
When AI creates ads from scratch (generating new concepts, layouts, and visual approaches):
When AI also designs the product packaging (removing even more creative constraints):
The pattern they found was clear and consistent: AI performs better with more creative freedom, not less.
The research team didn't just measure performance differences; they investigated why the gap exists. Three distinct patterns emerged:
In plain language: they're easier for consumers to process visually. Better color contrast between product and background. More balanced compositions. Clearer focal points. The ads feel effortless to look at, which translates directly to better engagement.
Think about scrolling through your social feed. Your brain makes split-second decisions about where to focus attention. Ads with higher processing fluency don't require conscious effort to understand. The visual hierarchy is clear. The product stands out appropriately. The composition guides your eye naturally.
AI models have identified the specific design patterns that reduce cognitive load for viewers: optimal contrast ratios between foreground and background, compositional balance that guides eye movement without effort, focal point placement that draws attention naturally. Human designers understand these principles too, but they're balancing them against dozens of other considerations: brand legacy, creative expression, stakeholder preferences, production constraints, etc. AI can generate designs optimized purely for visual clarity and cognitive ease.
The result is advertising that your brain can process with minimal effort, which the research shows drives measurably higher engagement.
This finding might be the most counterintuitive. You'd expect human creativity to be superior at emotional resonance. But the research found that AI-created ads scored significantly higher on emotional engagement than both human-created and AI-modified ads. The effect was even stronger when AI also designed the product packaging.
Why does this happen? AI models trained on millions of advertising examples can identify visual patterns and combinations that trigger emotional responses, even when those combinations aren't obvious to human designers. The models aren't "feeling" emotions, but they're recognizing what combinations of color, composition, context, and product presentation correlate with emotional engagement in the training data.
Here's what this means practically: AI can systematically explore emotional territories that human designers might never consider. A human creative director might default to established emotional cues: warm lighting for comfort, dramatic angles for excitement, natural settings for authenticity. AI doesn't have those same conventions limiting its exploration. It can test unexpected combinations and discover what actually drives emotional response with your specific audience.
The packaging finding is particularly revealing. When AI designed both the ad and the product packaging (the lowest creative constraint condition), emotional engagement increased even more. This suggests that emotional resonance isn't just about individual creative elements. It's about holistic visual coherence across all elements working together.
While AI-modified ads showed high novelty scores, they failed to preserve the authentic feel that makes ads effective and underperformed human-created assets by significant margins.
The modifications created subtle inconsistencies: elements that technically work individually, but feel slightly off in combination.
A shadow might fall at an unnatural angle. A product reflection might not match the lighting exactly. Brand elements might be subtly distorted. None of these issues would fail a quality check individually, but collectively they create a subconscious sense that something isn't quite right.
When you constrain AI to work within an existing human design, you're forcing it to navigate contradictions between what the original designer intended and what the AI model thinks performs well. The result isn't the best of both approaches. It's an ineffective compromise of the two.
When you let AI generate from scratch, it creates internally consistent designs optimized for visual processing patterns AND emotional engagement patterns the model learned from millions of advertising examples.
The research also uncovered another crucial factor: disclosure transparency.
When consumers knew AI was involved in creating the ads, advertising effectiveness actually decreased significantly. In some cases, click-through rates dropped by 32%.
This presents a challenging paradox for marketers: while AI-generated ads created from scratch outperform human-created ads when AI involvement is undisclosed, transparency about AI usage substantially undermines this advantage.
The disclosure penalty is particularly relevant given increasing regulatory requirements for AI disclosure, suggesting brands must carefully navigate the trade-off between compliance obligations and performance optimization.
While the research doesn't resolve this tension, it does help quantify it, which helps you make more informed decisions. Ultimately, when and how to deploy AI ads depends on many factors unique to your brand, and every brand’s approach will be different.
For CMOs and marketing technology leaders, these findings have direct operational implications. The most common approaches to AI creative deployment are not just suboptimal—they may actively hurt performance.
The most common approaches to AI creative deployment, like using it for A/B testing variations or minor optimizations, are not just suboptimal. They may actively hurt performance.
The organizations unlocking the greatest benefit from AI-generated ads are the ones architecting their creative operations to give AI systems the inputs, context, and freedom to generate original concepts from the ground up.
This doesn’t mean abandoning human creativity. Instead, it means redefining the human role in the creative process. Rather than using AI to polish human ideas, the most successful approach positions humans as creative directors who establish brand guidelines, strategic objectives, and quality controls while allowing AI to explore the full creative space within those parameters.
Human expertise becomes most valuable in curation, brand alignment, and upfront strategic direction rather than pixel-level execution.
At Adora, we've built our solution around this fundamental insight: AI delivers transformative results when empowered to create, not just iterate.
Our human-in-the-loop AI-powered creative optimization platform gives models the freedom to generate diverse, high-performing creative variations while maintaining brand consistency through strategic guardrails.
Rather than constraining AI to work within existing creative assets, we provide it with rich brand context, performance data, and strategic objectives then allow it to generate entirely new creative concepts. This approach has enabled customers like Alaska Airlines and Brooks Running to achieve the kind of performance improvements the research validates, not by asking AI to marginally improve existing assets, but by leveraging it as a creative force that can generate entirely new visual narratives at scale.
For marketing leaders, the question is no longer whether to adopt AI for creative production, but how quickly can you restructure your creative operations to capitalize on AI's proven performance advantages. If you’re only using AI to create variants or minor optimizations of existing assets, you’re competing with one hand tied behind your back.
The beauty industry participants in the NYU study discovered what we've seen across verticals: when AI is given the freedom to create rather than merely modify, it consistently discovers high-performing creative territories that human teams might never explore.
By combining this creative freedom with rigorous performance measurement and human oversight for brand compliance, enterprises can achieve the holy grail of digital advertising: creativity that's both breakthrough and measurably effective.
