AI adoption is booming, but productivity gains take time

Despite the hype, AI isn’t delivering the productivity gains many expected, according to a new study.

“Adoption is not a problem,” says Ian Baer, founder & CEO of AI-driven marketing insights company Sooth. “Quite the opposite: We have people adopting technology at a historic rate and applying it in ways that don’t positively impact their businesses. They’re checking the box for doing but not solving.”

A working paper from the National Bureau of Economic Research reinforces that point. While 70 percent of the firms surveyed reported using AI, 80 percent say it has no measurable impact on productivity. 

Even over the next three years, expectations are modest: a 1.4 percent increase in productivity and a 0.8 percent increase in output — marginal at best. 

For the advertising industry, the implications are significant. Adoption is already widespread, from personalization to performance tools to creative production. But the benefits are uneven. Just over half of respondents say AI saves time; fewer say it helps them scale.

This dynamic follows what economists call the “productivity J-curve.” As general-purpose technologies from electricity to the internet evolve, adoption comes fast, but actual productivity gains are slow.

“That can take years, not quarters,” says Baer. “And it also takes the courage to say you’re going to do it right, sometimes at the expense of doing it first. The smartest practitioners I know in brand environments have been piloting use cases for years before rolling out the performers.” 

Too many companies are using AI to accelerate production without questioning whether the underlying works.

“If you’re targeting the wrong audience with the wrong message on the wrong channel, AI just helps you waste money at scale,” Baer explains. “Like a robot taking your life savings into a casino and playing the law of averages at every seat at every table at the same time.”

“That’s what I’m seeing across the industry. AI is being bolted onto broken processes instead of being used to rethink the process itself.” 

Others point to a similar shift underway — from short-term efficiency to long-term capability.

“Right now, many companies are focused on short-term optics like efficiency,” said Angela Tangas, global CEO of OLIVER. “But the real opportunity is embedding AI in ways that actually drive growth: business, brand, and, of course, customers.”

“Building capability inside the brand is the key to staying one step ahead — and that means human talent.”

In practice, that means rethinking how work gets done, not just speeding it up.

“Brands require an operating model where talent understands brand, workflow, policy, stakeholders, customer — the full ecosystem — so that they can carry AI forward,” Tangas describes. “The human layer is what drives all the intelligence, craft, judgment and expertise to help make sure brands get results.”

She cited examples where AI is delivering real gains — particularly when it’s applied with clear intent. In one case, integrating generative AI into content production reduced non-working media costs, with savings reinvested into working media.

“I can safely say that we would not have been able to go as fast and launch it as quickly without AI,” says Jarie Bolander, the general manager and executive partner at Decision Counsel. “If you kind of look at the places where we’ve used it, the productivity gains have been five, 10, sometimes 20 times faster.”

But even in those cases, the role of human judgment remains central.

“Not everything we do should be automated with AI,” Bolander pointed out. “There are some things that are just not good, like the thoughtful analysis of what does ‘good’ look like? What’s the next best move?”

Part of what makes those efforts successful is the groundwork behind them.

“[Other companies] probably haven’t really taken a step back and said, ‘Okay, what are we automating?’” he explained. “‘How can we really architect our business processes?’ Because again, most places I’ve ever worked in, no one thinks from a systems perspective.”

Without that shift, many organizations risk getting stuck in what Tangas calls “pilot purgatory” — constantly experimenting with AI but not redesigning the production workflow.

“Yes, we are seeing gains in how quickly content is produced,” she said. “But the real organizational productivity gains are still a work in progress because they depend on how quickly people move up the learning curve.”

That puts pressure not just on technology investment, but on talent development.

“If agencies stop hiring junior strategists, junior analysts, junior creatives because AI can do the entry-level work, who becomes the senior talent in five years?” Baer said.

“You can’t AI your way to wisdom. You need people who’ve made mistakes, been wrong, and learned what works by doing the work.”