Marketing executives have varied views on generative AI, from fervent advocates to worried skeptics, but many are converging on a consensus: this technology will become "less sexy" in 2025. That is not necessarily a bad thing.

After a year offierce criticism of ads made with generative AI, the industry is beginning to focus more on back-office functions where automation tools can boost efficiency, scale operations, and avoid triggering consumer backlash. Evaluating key points of divergence in the customer journey or experimenting with synthetic audience data may yield more meaningful results than testing the latest software like OpenAI's Sora, at least in the short term. In the coming months, demands for concrete outcomes will also begin to separate winners from losers in the AI space, while emerging disruptors like DeepSeek's R1 model are increasing pressure to cut costs.

"Will AI start making TV ads on its own? That's an interesting topic for debate over drinks," said Josh Campo, CEO of digital agency Razorfish. "In the real world, at the stage we're at now, many things generative AI can do bring huge productivity gains. What it does best seems to be exactly the things people don't like doing."

Generative AI underpins a tension that has long plagued marketing decision-makers: keeping up with the latest tech trends while avoiding the "shiny object syndrome." In recent years, this trap has indeed appeared multiple times, as marketers bet on trends like the metaverse as the next big thing, only to quickly abandon those ambitions,leaving behind a digital wasteland。

AI feels different in its transformative potential, andthe scale of investment pouring into the fieldwill ensure its momentum lasts for some time. President Donald Trumpquickly reversed his predecessor's AI regulatory initiatives, while making the technology a key part of his administration's infrastructure agenda. Despite this windfall, in 2025 simply relabeling products as "now with AI" will not pass muster, and if the technology fails to meet inflated expectations, it could even become a deal-breaker. China-based DeepSeek, which delivered high-quality output at extremely low development costs within just a few days, shook up America's AI leaders—an attractive prospect for large enterprises with tight budgets.

"I don't think clients care whether you have AI if it doesn't produce the results they want. This is the year the industry really needs to deliver on its promises," said Lindsey DiGiorgio, CMO of advertising platform Yieldmo.

Growing unease

In 2024, consumer enthusiasm for generative AI waned as ads made with the technology or touting its benefits were repeatedly mocked. Tech giants like Google and Apple evenpulled ads that triggered dystopian alarms among viewers, while Coca-Cola ended the year with a particularly controversial holiday campaign.

"What AI does best seems to be exactly the things people don't like doing."

—Josh Campo, CEO of Razorfish

These attempts came from some of the most well-resourced, technologically capable, and established brand marketers, yet those credentials seemed to matter little. Althoughsome brands remain optimistic about AI-generated ad creative, the lukewarm reception in 2024 may set the tone for the coming months—a trend already reflected in recent consumer research.

According to a study released by NielsenIQ (NIQ) in December, consumers consistently rated AI-generated video ads as more "annoying," "boring," and "confusing" than traditional ads. Even AI-generated content deemed high-quality failed to make a strong impression on respondents, with viewers subconsciously sensing something was off.

The "uncanny valley" problem may have heightened concerns about letting AI lead creative projects. According to a report from Yieldmo and Ascendant Network, more than a third (38%) of marketers feel generally uncomfortable applying generative AI to any scaled marketing campaign. More marketers may lean toward a "hybrid" approach, using AI to enhance existing assets without letting the technology take center stage.

"I do think the use of generative AI will continue to be hidden," said Chris Neff, head of global emerging experiences and technology at creative agency Anomaly. "Methods will be blended. And with blending, costs may also be lower."

Marketing leaders also worry that AI's ability to instantly generate large volumes of similar content could dilute brand distinctiveness at a time when customer loyalty is critical. Craig Brommers, CMO of American Eagle, recently spoke at anindustry eventpanel about his concern that the technology could lead to "generic creative." He further noted this could impact the authenticity of brands like Aerie that prioritize diversity and inclusion. This worry is shared by others in the industry who fear AI could amplify human biases.

"There's also a real diversity and inclusion aspect to this story. AI draws from everything that exists and doesn't always represent different groups best," said Megan Belden, vice president of Bases Advertising at NIQ and one of the authors of the AI advertising study.

Boosting efficiency

Generative AI may still be immature in delivering final creative products, but its influence in other aspects of the production process will rise in 2025. Early-stage tasks like briefs, research, and storyboards may get a boost as marketers face pressure to further compress budgets.

"What's currently undervalued is the process part," said Lance Wolder, head of strategy at PadSquad.

Adapting assets for localization or different media formats is a promising area, and major tech platforms like Meta, Amazon, and Adobehave already tried to capitalize on this opportunity through expanding AI product lines. For example, Adobe Firefly'sBulk Create featurecan edit thousands of images in one click, swapping different backgrounds and resizing, potentially saving dozens of hours of manual labor.

"There's a lot of work currently dedicated to adaptation," said Campo of Razorfish. "If I can process data and adapt more assets faster, I should also be able to deliver more personalized experiences to consumers."

"The biggest area of growth—and there's already been a lot—is in audience targeting."

—Lindsey DiGiorgio, CMO of Yieldmo

Targeting and campaign optimization may also benefit from the latest AI capabilities, which can identify patterns in large datasets and provide synthetic audiences for testing ads. Improving precision remains a top priority,as brands try to move away from reliance on third-party cookies, even though Google last year pulled back on its plan to phase out that targeting technology.

"The biggest area of growth—and there's already been a lot—is in audience targeting," said DiGiorgio. "These applications are what people are most familiar with because they involve large, heavy datasets as inputs."

According to research from the Interactive Advertising Bureau, eight in ten media buyers are exploring generative AI to some degree, but only a third have organized collaborative resources around the technology. If marketers want to achieve productivity gains rather than being overwhelmed by the sheer volume of options the technology can generate, organizational readiness around AI is crucial.

"Just because you can generate tons of iterations to tell a story or provide a brand team with a huge number of assets doesn't mean you should," said Wolder of PadSquad. "The key is how we use these tools to improve our own efficiency."

A glut of choices

As marketers ponder how best to deploy generative AI, the number of platforms and partners available to them has surged. "Overwhelming" was a descriptor repeated by several experts, who expect some shakeout in 2025 as companies creating industry-specific and comprehensive generative AI products stand out. Platforms transparent about how their AI models are trained may also win over risk-averse marketers.

"No one wants to be the first big brand to get sued."

—Chris Neff, Head of Global Emerging Experiences and Technology at Anomaly

"Over the next 12 months, filtering out the flashy from the genuinely useful will be the hardest part, because the rush of solutions being offered right now is so massive," said Wolder.

Companies poised to benefit from the generative AI wave include digital advertising giants, which leverage their scale and maturity to serve performance-focused brands with smaller budgets that either can't afford more expensive models or are more focused on producing assets in volume. In Q3 2024, over one million advertisers used Meta's generative AI tools, creating more than 15 million ads in a single month. Amazon has alsomade similar progress in guiding merchants toward its AI tools, which now include video, audio, text, and image generators.

"You'll see small businesses rally around these tools because they support them," said Neff of Anomaly, referring to interest in generative AI among small and mid-sized marketers. "It will help their margins and make them feel more powerful."

That said, DeepSeek's rapid rise shows how quickly the competitive landscape of generative AI can reshuffle. While some brands are reluctant to entrust sensitive data to a Chinese startup—which recentlysuffered a cyberattack—according to The Wall Street Journal,others are already exploring potential applications, eagerly anticipating that U.S. competitors will cut prices as a result. Consumers also seem excited about the development: according to Sensor Tower, DeepSeek quickly climbed to the top of app store charts, surpassing startups like Perplexity. The research firm said DeepSeek's app downloads have exceeded 3 million, with 80% occurring in the past week.

Excitement mixed with anxiety over software like DeepSeek encapsulates the double-edged sword generative AI poses for many brands. For large marketers that protect intellectual property and stand to lose more from missteps, handing valuable materials to large language models and machine learning algorithms remains a daunting prospect, regardless of the platform.

"Ownership is the biggest issue, and frankly, it's still unclear," said Wolder. "We're talking about a billion-dollar brand potentially leaking its brand details, its plan details. It's a challenging time, and I think some legal departments are justified in ensuring the sandbox or environment where these tools operate is truly protected."

Conversely, the idea that AI models might generate assets plagiarizing content from other artists or companies is an ongoing concern in advertising and beyond. Neff of Anomaly summed up the dilemma in a few sentences:

"No one wants to be the first big brand to get sued."