Marketing executives hold a range of opinions about generative artificial intelligence (AI), from the evangelistic to the alarmist, but many are aligning around a shared sentiment: The tech is getting “less sexy” in 2025. That’s not necessarily a bad thing.
Following a year during which campaigns made with generative AI received a shellacking, greater attention is being paid to the back-of-house functions for which automation tools can improve efficiency, provide scale and steer clear of consumer outcry. Assessing forks in the customer journey or toying with synthetic audience data could drive more meaningful results than experiments with the latest software like OpenAI’s Sora, at least in the near term. In the months ahead, demands for concrete outcomes will also start to sort out AI’s winners and losers, with emergent disruptors like DeepSeek’s R1 model intensifying pressure to lower costs.
“Is [AI] going to just start creating TV ads on its own? That’s a fun topic to debate over drinks,” said Josh Campo, CEO of digital agency Razorfish. “In the real world, where we’re at today, there’s a lot of things that gen AI can do that have dramatic productivity impacts. The things it does best seem to be the things that people don’t like to do.”
Generative AI underpins a tension that has long dogged marketing decision-makers: the need to keep up with the latest technology trends without falling victim to shiny penny syndrome. The shiny penny trap has indeed been sprung several times in recent years, with marketers making crazes like the metaverse their next big bet before quickly abandoning those ambitions, leaving digital wastelands in their wake.
AI feels different in regard to its transformational potential, and the amount of investment pouring into the category will ensure momentum for some time to come. President Donald Trump has moved fast to peel back his predecessor’s AI oversight initiatives while making the technology a key piece of the administration’s infrastructure agenda. Despite the windfall, simply relabeling a product as “Now with AI” won’t pass muster in 2025 and could even be a dealbreaker if the tech doesn’t measure up to lofty expectations. China-based DeepSeek in just a few days time has rattled the sector’s U.S. vanguard by delivering quality outputs at a fraction of the development cost — an appealing prospect for large enterprises working with tightened belts.
“I don’t think that clients care if you have AI if it doesn’t produce the results that they like. This is the year that the industry really has to pay that off,” said Lindsey DiGiorgio, CMO of advertising platform Yieldmo.
Growing unease
Consumer enthusiasm for generative AI deflated in 2024 as ads made with the technology or promoting its benefits were repeatedly subject to derision. Tech giants like Google and Apple went as far as to pull commercials that sounded dystopian alarms for viewers while Coca-Cola capped off the year with a particularly contentious holiday campaign.
“The things [AI] does best seem to be the things that people don’t like to do.”
Josh Campo
CEO, Razorfish
These efforts were made by some of the most well-resourced, technologically capable and storied brand marketers, but the bona fides didn’t seem to matter. While some brands will stay bullish on AI-generated advertising creative, the chilly reception in 2024 could set a tone for the months ahead — one that’s been reflected in recent consumer research.
Consumers consistently rated AI-generated video advertising more “annoying,” “boring” and “confusing” than conventional ads, per a study published by NielsenIQ (NIQ) in December. Even AI-generated content perceived as high quality did not create as strong an impression in people polled by the group, with viewers subconsciously reading something amiss.
The uncanny valley problem could be contributing to growing unease at the idea of letting AI take the wheel on creative initiatives. Over one-third (38%) of marketers are roundly uncomfortable applying generative AI to any scaled marketing campaigns, according to a report by Yieldmo and Ascendant Network. More marketers may gravitate toward a “blended” approach, using AI to augment existing footage without making the tech the star of the show.
“I do believe just the use of generative AI will continue to get hidden,” said Chris Neff, global head of emerging experience and technology at creative agency Anomaly. “The methods will be blended. By blending, it also can be cheaper.”
Marketing leaders have also worried that AI, which can produce reams of similar-looking content in an instant, risks diluting brand distinctiveness at a time when customer loyalty is paramount. American Eagle CMO Craig Brommers spoke of his concerns that the technology results in “generic creative,” during a recent trade show panel. This could affect the authenticity of brands like Aerie that prioritize diversity and inclusion, he went on to say. It’s a trepidation shared by others in the industry who are wary of AI exacerbating human biases.
“There’s a real diversity and inclusion aspect also to this story. AI is pulling from everything that exists and isn’t always able to do the best job of representation,” said Megan Belden, vice president for NIQ’s Bases Advertising, an author of the AI in advertising research.
Fueling efficiency
Generative AI may still be underbaked when it comes to realizing a final creative product, but its influence over other aspects of the production pipeline will climb in 2025. Early stage tasks like briefing, research and storyboarding could see a boost as marketers are challenged to stretch their dollars further.
“The underappreciated thing right now is the process part,” said Lance Wolder, head of strategy at PadSquad.
Adjusting assets for localization purposes or different media formats is an area of promise and one which large technology platforms like Meta, Amazon and Adobe have tried to capitalize on with an expanding slate of AI products. For example, Adobe Firefly’s recently announced a Bulk Create feature that can edit thousands of images in a single click, swapping in different backgrounds and resizing images, potentially saving dozens of hours in manual labor.
“There’s a lot of work that gets done today specifically around adaptation,” said Razorfish’s Campo. “If I can process the data, and I can adapt more assets more quickly, I should also be driving a much more personalized experience for consumers.”
“The biggest area where there’s going to be a lot of growth — there is already a lot of growth — is in that audience-targeting area,”
Lindsey DiGiorgio
CMO, Yieldmo
Targeting and optimizing campaigns could also get an assist from the latest AI bells and whistles, which can identify patterns in large data sets and provide synthetic audiences to test ads against. Improving precision remains top of mind as brands try to wean off a reliance on third-party cookies despite Google backtracking on its plans to wind down the targeting technology last year.
“The biggest area where there’s going to be a lot of growth — there is already a lot of growth — is in that audience-targeting area,” said DiGiorgio. “Those are the applications that people are most comfortable with just because they involve large, heavy datasets as inputs.”
Eight in 10 media buyers are exploring generative AI to some degree, per research from the Interactive Advertising Bureau, but just one-third have organized, collaborative resources around the technology. Marketers getting their ducks in a row around AI will be essential if they want to realize productivity gains rather than getting swamped in the swaths of options the technology can crank out.
“Just because you can generate tons of iterations to tell a story or to give your brand team a ton of assets to use doesn’t mean you should,” said PadSquad’s Wolder. “It’s about how we use the tools in ways to make ourselves more efficient.”
Swamped with choice
As marketers mull how to best deploy generative AI, the number of platforms and partners available to them has ballooned. “Overwhelming” was a descriptor repeated by several experts, who expect that 2025 will see some culling as companies creating industry-specific and well-rounded generative AI products rise to the top. Platforms that are transparent around how models AI are trained could also curry favor with risk-averse marketers.
“Nobody wants to be the big brand that’s sued first.”
Chris Neff
Global head of emerging experience and technology, Anomaly
“How you filter out the fluff versus the real thing is going to be the hardest part in the next 12 months as there’s such a gold rush to offer a solution,” said Wolder.
Those primed to reap the benefits of the generative AI wave include digital advertising’s heavyweights, which are leveraging their scale and sophistication to cater to the smaller-pocketed and performance-minded brands that either can’t afford costlier models or are more squarely focused on churning out assets. More than 1 million advertisers took advantage of Meta’s generative AI tools in Q3 2024, creating over 15 million ads over the course of a month. Amazon has touted similar progress in onboarding merchants to its AI tools, which now include video, audio, text and image generators.
“You’re going to see small businesses rally behind these tools because it’s going to prop them up,” said Anomaly’s Neff of small- and mid-sized marketers’ interest in generative AI. “It’s going to help their bottom line, it’s going to make them feel bigger.”
That said, the fast rise of DeepSeek speaks to how quickly the competitive field could be shuffled for generative AI. While some brands will be reticent to trust sensitive data to a China-based startup — one that has recently been subject to a cyberattack — others are already exploring potential applications and eagerly anticipating U.S. rivals will cut prices in response, The Wall Street Journal reported. Consumers similarly seemed enthused by the development: DeepSeek skyrocketed to the top of the app store charts, outpacing startups like Perplexity, according to Sensor Tower data. The DeepSeek app has surpassed 3 million downloads , 80% of which occurred in the past week alone, the researcher said.
Equal helpings of excitement and anxiety about software like DeepSeek encapsulates the double-edged sword generative AI poses to many brands. For larger marketers that are protective of their intellectual property and arguably have more to lose in the case of a stumble, forking over valuable material to large language models and machine-learning algorithms is still a daunting prospect regardless of the platform.
“Ownership is the biggest piece of this, and it’s still unclear, to be honest,” said Wolder. “We’re talking about a billion dollar brand losing potentially specifics about their brand, specifics about their plans. It’s a challenging time and I think some of these legal departments are justified in making sure that the sandbox or the world with which these tools operate is truly protected.”
Conversely, the idea that an AI model could spit out assets that crib from other artists or companies is a going concern, in advertising and elsewhere. Anomaly’s Neff summed up the conundrum in a few words:
“Nobody wants to be the big brand that’s sued first.”
