--- title: AI Creative Campaigns: When AI Drives Marketing Creativity description: Real examples of AI in creative campaigns. What worked, what backfired, and what these experiments teach about the future of marketing creativity. date: February 5, 2026 author: Robert Soares category: ai-use-cases --- Coca-Cola remade its beloved 1995 Christmas commercial with AI. The internet called it soulless. A fintech startup spent $2,000 on an AI ad that aired during the NBA Finals. It went viral for all the right reasons. Same technology. Opposite outcomes. The gap between these results isn't random, and understanding it matters more than any debate about whether AI belongs in creative work at all. Because that ship has sailed. [83% of ad executives now deploy AI in the creative process](https://mountain.com/blog/ai-generated-tv-commercials/), up from 60% just two years ago. The question now is how to use it without alienating the people you're trying to reach. ## The Coca-Cola Disaster In November 2024, [Coca-Cola unveiled an AI-generated remake of "Holidays Are Coming"](https://www.nbcnews.com/tech/innovation/coca-cola-causes-controversy-ai-made-ad-rcna180665), the iconic 1995 ad featuring red trucks rolling through snowy towns. Three AI studios and four different generative models produced the new version. The backlash was immediate. Critics called it "creepy," "dystopian," and "devoid of genuine creativity." The irony of ending with the tagline "Real Magic" while showing obviously synthetic imagery didn't go unnoticed. [Sentiment analysis from Marketing-Interactive](https://www.marketing-interactive.com/coca-cola-ai-remake-holiday-ad-mixed-sentiments) showed positive reactions dropped from 23.8% before the campaign launch to just 10.2% after. Negative sentiment held steady at around 32%. What went wrong? The original ad works because it captures something ineffable about holiday nostalgia. The warm glow of lights against snow. The anticipation of something arriving. AI can generate images of trucks and snowflakes. It struggles to generate the feeling those images are supposed to evoke. As [keiferski noted on Hacker News](https://news.ycombinator.com/item?id=40725329): "AI is being added to the name of every product. Not because it's actually AI in any rigorous or historical sense of the word, but because it's trendy and helps you get investment dollars." That observation cuts to something deeper about why the Coca-Cola ad failed. It felt like a technology demonstration, not a piece of communication designed to connect with humans. The studios involved defended the speed of production. What used to take twelve months took two. But nobody watching an ad cares how fast it was made. They care whether it makes them feel something. ## Toys R Us Tries to Make History A few months earlier, [Toys R Us premiered what they called the first brand film made with OpenAI's Sora video tool](https://www.marketingdive.com/news/toys-r-us-openai-sora-gen-ai-first-text-video/719797/). They showed it at Cannes Lions in June 2024. The one-minute video imagined how founder Charles Lazarus dreamed up the toy store as a child. The venue was a mistake before anyone pressed play. Cannes Lions attracts creative professionals who are watching AI encroach on their livelihoods. Premiering an AI-generated ad there, to that audience, was asking for trouble. The backlash [started immediately](https://www.nbcnews.com/tech/internet/toys-r-us-ai-video-ad-controversy-explained-commercial-rcna159030). People called it "cynical" and "soulless." The technology wasn't ready either. Characters' faces warped between shots. The child protagonist looked like a different person from one frame to the next. The chief creative officer acknowledged that getting facial expressions and motivations correct was "particularly challenging." Toys R Us declared the campaign successful anyway. They noted AI completed 80-85% of the work. But measuring success by efficiency ignores the entire point of brand advertising. The goal is connection. Efficiency that produces disconnection isn't a win. ## The Kalshi Ad Nobody Expected While major brands stumbled, a prediction market platform called Kalshi did something strange. [They spent $2,000 on an AI-generated ad](https://www.npr.org/2025/06/23/nx-s1-5432712/ai-video-ad-kalshi-advertising-nba-finals) and ran it during the NBA Finals. The ad featured a farmer submerged in a pool of eggs. An alien drinking beer. Bizarre characters throughout. It leaned hard into the surreal quality of AI-generated video instead of trying to hide it. P.J. Accetturo, who created the video, [told NPR](https://www.npr.org/2025/06/23/nx-s1-5432712/ai-video-ad-kalshi-advertising-nba-finals): "This took about 300-400 generations to get 15 usable clips." The process wasn't magic. It was iteration. He also added something important: "Just because this was cheap doesn't mean anyone can do it." Kalshi's Jack Such put the cost in perspective. "The actual cost of prompting the AI, what is being used in lieu of studios, directors, actors, etc., was under $2,000." The total production costs were higher once you factor in strategy, editing, and placement, but the creative production itself cost almost nothing compared to traditional advertising. So why did this ad work when Coca-Cola's didn't? Context. A scrappy fintech startup running a weird ad during basketball matches the brand. Nobody expected polish. The absurdism fit. The AI-ness was part of the joke rather than something to hide. Debra Aho Williamson, a marketing analyst at Sonata Insights, made an observation worth considering. When her firm [asked Gen-Z and millennial consumers](https://www.npr.org/2025/06/23/nx-s1-5432712/ai-video-ad-kalshi-advertising-nba-finals) how positively they felt about AI-generated ads, only 48% responded positively. That's less than half. The bar is high. ## Virgin Voyages Gets Personal [Virgin Voyages launched "Jen AI"](https://storychief.io/blog/ai-marketing-campaigns), an avatar inspired by Jennifer Lopez that sends personalized video invitations for special occasions. Each message is unique. Each feels like a playful one-on-one communication. The transparency is built into the name. It's called Jen AI. There's no pretense that Jennifer Lopez personally recorded a birthday greeting for you. The invitation is to enjoy something fun, not to believe something false. This works because AI enables something that wasn't possible before. The real Jennifer Lopez can't record personal messages for everyone who books a cruise. The AI version can. That's a genuine value proposition rather than a cost-cutting measure disguised as innovation. [Research from Smartly](https://www.smartly.io/resources/ai-and-advertising-in-2025-what-consumers-really-expect) found that 48% of consumers trust ads co-created by a person with AI support, compared to just 13% who trust ads created entirely by AI. The hybrid model, human direction with AI execution, lands better than pure automation. ## Burger King Flips the Script [Burger King's "Million Dollar Whopper" campaign](https://storychief.io/blog/ai-marketing-campaigns) turned customers into creators. People designed dream Whoppers online, picking toppings in whatever combinations they wanted. AI generated photorealistic images of their creations along with custom jingles. The ideas came from humans. AI visualized them. This inversion matters. Instead of AI making something for audiences to passively consume, AI helped audiences make something for themselves. The million-dollar prize created a reason to participate, but even non-winners got shareable digital assets of their creations. The stakes were appropriately low too. A slightly off AI rendering of a fantasy burger is fine. Nobody expects photorealism from an imaginary sandwich. The campaign didn't ask AI to be emotionally resonant. It asked AI to be fast and responsive. AI is good at fast and responsive. ## What the Patterns Reveal Alok Saboo, a marketing professor at Georgia State University, captured something essential in his assessment of the Kalshi ad. [He told NPR](https://www.npr.org/2025/06/23/nx-s1-5432712/ai-video-ad-kalshi-advertising-nba-finals): "In the end, humans want to connect with humans." That simple observation explains most of the variance in AI campaign outcomes. When AI enables human connection, campaigns work. When AI substitutes for human connection, campaigns fail. The successful campaigns share certain features. They're transparent about AI involvement. They use AI to do something new rather than to replicate something old more cheaply. They keep humans in the creative loop, letting AI scale ideas rather than generate them. They match AI aesthetics to brand positioning, using weirdness deliberately rather than accidentally. The failures share features too. They try to recreate emotional experiences with synthetic tools. They position efficiency as the primary value. They hide or minimize AI involvement, then face backlash when audiences figure it out anyway. They use beloved source material, inviting unfavorable comparisons. Claire Xue, an AI creative consultant who has worked with LVMH and Sephora, [identified the resistance from major brands](https://www.ibm.com/think/news/ai-generated-advertising-2025): "We're seeing pushback, especially from bigger companies, over concerns about maintaining brand standards and avoiding public backlash or intellectual property issues." Those concerns are valid. The question isn't whether to use AI. It's whether your specific use case will help or hurt. ## The Authenticity Paradox Here's something counterintuitive. Audiences seem to forgive obvious AI more readily than they forgive AI pretending to be human. Kalshi's bizarre imagery worked partly because nobody could mistake it for traditional filmmaking. The uncanny quality was the aesthetic. Virgin Voyages' Jen AI works because the name announces what it is. Burger King's visualizations work because imaginary food is obviously imaginary. Coca-Cola's ad failed partly because it tried to replicate something that felt human with tools that couldn't deliver humanity. The attempt to hide limitations made them more glaring. As [cleandreams observed on Hacker News](https://news.ycombinator.com/item?id=40725329) about AI broadly: "The tech is good enough to make incredible demos but not good enough to generalize into reliable tools. The gulf between demo and useful tool is much wider than we thought." That observation applies directly to AI creative campaigns. Demo mode and campaign mode are different things. What looks impressive in a controlled presentation may look hollow in the wild. ## Consumer Research Cuts Both Ways The data on consumer attitudes toward AI advertising is mixed, which is actually useful information. [Research from NIQ via Marketing Dive](https://www.marketingdive.com/news/consumer-perceptions-generative-ai-in-marketing-openai-sora/735761/) found that AI-generated creative is consistently assessed as more "annoying," "boring," and "confusing" than traditionally produced ads. Even AI ads perceived as high quality didn't leave as memorable an impression. But that same research landscape includes examples of AI campaigns that performed well. The difference appears to be whether AI serves a clear purpose beyond cost reduction. Joe Prota, Director of Brand Marketing at IBM, [described their internal AI creative tools](https://www.ibm.com/think/news/ai-generated-advertising-2025): "Our team was able to prompt Firefly to create a fish that looked like a hamster, and they achieved it in hours instead of days or weeks." That speed matters for iteration. He added important nuance though: "With content like emails and social media posts, AI can generate multiple versions effectively. But for larger campaigns, human input remains critical." The pattern holds across examples. AI for variants, testing, and personalization works. AI for hero creative trying to carry emotional weight struggles. ## Where This Is Heading The brands doubling down on AI aren't abandoning it despite the failures. They're getting more selective about application. [H&M announced plans to use AI-generated digital twins of real models](https://www.superside.com/blog/ai-marketing-campaigns) for some marketing assets. The reception was mixed. It solves a production problem (generating diverse assets quickly) while raising questions about authenticity and job displacement. Julien Vallée, a visual artist and commercial director, [noticed a shift in how agencies brief him](https://www.ibm.com/think/news/ai-generated-advertising-2025): "Most briefs we get from agencies now come with a clear vision created using tools like Midjourney." He appreciates the efficiency but warns about "unrealistic expectations" that AI visuals can create. What's easy to generate isn't necessarily easy to execute. The industry seems to be arriving at a consensus. AI tools are part of the creative toolkit now. The question is which tool for which job. A hammer is useful. That doesn't make everything a nail. Tom Greenhalgh, Data & Measurement Lead at Google, [summarized the data angle](https://www.thedrum.com/news/2024/12/19/ai-roi-lessons-2024-will-set-marketers-up-success-2025): "AI, when fed with high-quality first-party data, can help us understand customer behaviors faster than ever before." The insight isn't about replacing creative judgment. It's about informing it. ## Questions Worth Asking Before launching an AI creative campaign, some questions help predict outcomes. Does AI enable something new, or does it replicate something old more cheaply? Audiences can tell the difference. New things get curiosity. Cheap replicas get contempt. Is the AI involvement transparent? Hidden AI triggers backlash when discovered. Announced AI sets appropriate expectations. Does the brand context fit AI aesthetics? Challengers and tech brands have more latitude. Heritage brands and emotional categories have less. Are humans directing the creative, with AI executing? Or is AI generating, with humans just approving? The direction of the relationship matters. What happens if this fails publicly? Coca-Cola can survive a bad ad. A smaller brand might not recover as easily from becoming the example of what not to do. The technology will keep improving. What's uncanny today may feel natural in two years. But right now, the gap between AI capability and audience expectation remains significant. The brands navigating that gap successfully are the ones treating AI as a tool rather than a replacement for the judgment about when and how to use it. That judgment is still human. For now.