For years, the advertising industry has treated the arrival of AI-generated creative as a looming disruption rather than a proven reality. Marketers debated whether machine-made visuals and copy could ever match the intuition, cultural sensitivity, and emotional intelligence that experienced human creatives bring to a brief. In 2025 and into 2026, the debate finally has real data behind it — and the answers are more nuanced, more instructive, and arguably more consequential than either side of the argument anticipated.
The most rigorous evidence to date comes from a landmark field study published in January 2026, conducted in collaboration with researchers from Columbia University, Harvard University, the Technical University of Munich, and Carnegie Mellon University. Using live campaign data from Taboola’s performance advertising platform, Realize, the study examined hundreds of thousands of advertisements across more than 500 million impressions and three million clicks — making it one of the largest real-world comparisons of AI and human creative performance ever conducted. The findings were striking: AI-generated ads performed on par with human-made ads, and in raw numbers, actually edged ahead. AI creatives recorded an average click-through rate of 0.76 per cent against 0.65 per cent for human-made ads — a gap that narrowed under tighter statistical controls but remained directionally consistent throughout the analysis.
What the headline numbers do not tell you, however, is the story behind the story. The study’s most revealing finding was not that AI won. It was why AI won — and when it did not. AI-generated ads that did not look like AI achieved the highest engagement of all groups, significantly outperforming both human-made ads and AI ads that were visibly artificial. The implication is direct: the “AI vs human” framing may be the wrong lens entirely. What audiences respond to is not who — or what — made an ad, but whether that ad feels authentic, trustworthy, and human in its emotional register. The technology is only an advantage when it disappears.
The Taboola study also identified a specific visual element that functioned as a reliable trust signal across both AI and human creatives: the presence of a large, clear human face. Interestingly, AI-generated ads were more likely to include prominent human faces than their human-made counterparts, which may partly explain the raw performance advantage. The implication for creative teams is meaningful — not that AI should replace photographers or art directors, but that the instincts built into AI systems, trained on high-performing creative libraries, have begun internalising lessons about visual communication that practitioners sometimes deprioritise under production pressure.
“AI-created ads can even outperform human-created ads, which serves as a good reminder of AI’s incredible power and scale when harnessed intelligently — but it needs human guidance to create outputs that feel authentic.” — Taboola
The cost dimension adds another layer of significance to these findings. AI-generated creatives can be produced at a fraction of what a traditional production process costs — estimated at just a few cents per asset request — and at a speed that fundamentally alters testing economics. Where traditional A/B testing required weeks of planning, production, and data accumulation, AI now compresses that cycle. Platforms using AI-powered creative tools have reported creative testing happening up to ten times faster, enabling marketers to run parallel experiments across headlines, imagery, colour palettes, and calls-to-action that would have been logistically impossible in a conventional workflow. The result is not simply efficiency — it is a higher probability of finding what actually works, because more hypotheses can be tested in the same window of time.
The performance picture, though, is not uniformly positive for AI. A critical distinction emerges when the metric shifts from direct-response to brand-building outcomes. Nielsen’s 2025 study on advertising effectiveness found that human-crafted brand campaigns generated 43 per cent higher unaided recall and 37 per cent higher emotional engagement scores compared to AI-generated equivalents. This divergence is not accidental. Brand storytelling demands something that current AI systems struggle to synthesise: a coherent point of view, cultural specificity, and the ability to make an audience feel something that outlasts the impression. AI can optimise for clicks. It cannot yet reliably optimise for meaning.
Kantar’s testing of AI-generated advertising adds a further complication. Facial coding data from their research revealed that while GenAI ads provoke stronger emotional reactions overall, they also skew negative more often than human-made ads. The “uncanny valley” effect — that unsettling quality of something that looks almost human but not quite — remains a live risk in AI visual production. An ad that triggers discomfort, confusion, or amusement at its artificiality is generating emotional engagement in the wrong direction. Kantar also found that ads where AI integration was seamless — where the viewer had no reason to notice the technology — performed in the top tier for branded cut-through, with over 40 per cent landing among the best-performing creative in their category. The lesson is not that AI creatives underperform, but that detectable AI creatives do.
The video format reveals an additional dimension of the performance gap. For video ads longer than 15 seconds, human creatives deliver 28 per cent higher completion rates and 19 per cent higher click-through rates. The gap narrows considerably for short-form content — sub-six-second bumper ads, for instance, show comparable performance between AI and human production. This is not surprising when you consider what longer video demands: narrative arc, pacing, tonal consistency, and an understanding of how emotional states build across time. These are precisely the capabilities where human creative direction continues to hold a structural advantage, and where AI functions better as an executor than an originator.
India’s advertising ecosystem offers a particularly instructive lens on these global findings. As a market with intense emotional investment in culturally resonant storytelling — where campaigns like Google India’s Reunion ad, Tanishq’s wedding films, and Cadbury Dairy Milk’s generational narratives have defined what advertising can accomplish — the limits of AI become more visible. Research consistently shows that Indian consumers respond to advertising through the emotional circuits of nostalgia, humour, and familial bonding. These are not patterns that can be reverse-engineered from performance data alone. By the end of 2025, Indian advertising was operating within a structurally changed ecosystem, with AI transitioning from experimentation to infrastructure — yet even within this shift, the most culturally significant work continued to emerge from human-led creative thinking.
The hybrid model that has emerged from all of this evidence is neither a compromise nor a concession. It is, increasingly, the competitive standard. Analysis of campaign performance across major advertisers shows that hybrid workflows — where humans lead strategic direction and AI handles variation generation, localization, and testing — outperform either approach alone by over 30 per cent. Platforms like Smartly.io have demonstrated this at scale: a single human-conceived template, when processed through AI for localization across 20 or more markets, can generate thousands of culturally adapted variants in hours. The creative intelligence is human. The production leverage is machine.
What this means practically for creative and marketing teams is a reconfiguration of roles rather than a reduction of them. The fastest-growing position in advertising agencies in 2025 and 2026 is the Creative Strategist — a hybrid profile that combines data fluency with creative judgment, interprets performance signals from AI testing, and translates those signals into the next round of strategic concepts. LinkedIn data showed a 340 per cent increase in job postings for this role since 2024. These are not people who design ads manually or who feed prompts into image generators. They set the creative direction that makes AI output worth testing in the first place.
The upshot of all this testing is a reframing of the original question. A/B testing did not reveal that AI is better than humans or that humans are safer than AI. It revealed that the source of an ad matters far less to an audience than how that ad makes them feel, and that the most effective creative teams are those who have stopped treating the two as competitors. AI excels at volume, variation, speed, and the identification of patterns within performance data. Humans excel at cultural intuition, long-form narrative, emotional authenticity, and the strategic judgment that determines what should be made in the first place. Neither set of strengths is dispensable. The brands that have recognised this earliest — and built workflows accordingly — are the ones whose A/B tests keep returning results worth acting on.
