Managers Use AI More, Like It Less: Wharton Survey Shows Split

By Daniel Okafor

Sixty-two percent of U.S. managers and above now use AI daily, up from 46% a year earlier, while the share calling themselves excited fell eight percentage points to 51%. That split comes from the new annual survey by Wharton Human-AI Research and research firm GBK Collective, and the researchers call it a first: in four years of the study, usage and excitement had never moved in opposite directions.

"The amazement is kind of wearing off a little bit," Prasanna "Sonny" Tambe, professor of operations, information and decisions and co-director of Wharton Human-AI Research, told Fortune. "This technology is moving so fast and our organization is not ready to keep up."

The cooling is not a backlash. Optimism among managers and above held steady at 59%, and 80% reported positive financial returns from AI — though only 28% called those returns strongly positive. A quarter described themselves as skeptical; 44% as cautious.

What does the shift look like on the ground?

The managers' mood now matches what younger users already report. A Gallup survey published in April, conducted with the Walton Family Foundation and GSV, found 51% of Americans aged 14 to 29 used generative AI at least weekly, even as excitement fell 14 points year over year. Among employed Gen Z respondents, 48% said AI's workplace risks outweighed its benefits, even as 56% said the tools help them work faster.

For Jeremy Korst — a Wharton lecturer, GBK partner and founder of AI consultancy Mindspan Labs — the cooling reflects the move from experimenting with a technology to making it work inside an organization. He compared it to going from an "Oh, wow" level of excitement to working out how the transition actually happens.

The most consequential change may come after AI produces an answer. Seventy-two percent of managers and above said their focus had shifted from creating outputs to managing AI-generated outputs. The figure was 63% among managers and directors, and 82% among vice presidents and above.

That shift puts verification at the center of the job. "I think we have been overindexing a little bit on what agents can do, as opposed to … what is it that we can verify that agents can do," said Tambe, who is on sabbatical and working on economic research from within OpenAI. Verification gets harder, he said, when an agent's work crosses organizational functions.

Who answers for an AI mistake?

Organizations have not settled who is accountable when autonomous AI makes a consequential error. The survey found:

• 28% of managers and above named a governance or risk team
• 26% named the deploying employee's manager
• 25% named the employee
• Among managers and directors, one in five said nobody was clearly accountable, or they did not know. Among VPs and above, 3% said the same.

"It's a simple question, but from an organizational perspective, it's pretty hard because it involves job roles and complicated things like that that actually take time to resolve," Tambe said.

Korst pointed to research by Wharton's Gideon Nave and Steven Shaw on "cognitive surrender" — whether reviewers truly check AI's work or simply sign off at some point.

For now, most departments keep a person in the loop. Among managers who identified their department's highest permitted AI autonomy level, 48% said AI performs work but a human approves it before execution. Another 35% said AI only provides suggestions or drafts. Just 17% reported autonomous operation with after-the-fact sampling or real-time monitoring and guardrails.

Where is the strain landing?

On the middle of the organization. Korst said the shift is "putting a bunch of strain in the middle of the organization," describing "middle managers and teams who are tasked to lead this transformation and do it right." That layer, he added, "doesn't have a lot of capacity."

The strain runs downhill. A third of managers and above said their departments had already slowed or frozen hiring because of AI. 24% expect fewer entry-level hires over the next two to five years, against 14% who expect fewer executive hires.

Bosses and employees also disagree about readiness: 78% of managers and above said employees receive adequate AI training, but only 56% of entry- and mid-level employees agreed. Junior employees joined the survey for the first time this year; the data do not match managers and employees within the same companies.

Who is actually getting returns?

The departments seeing strongly positive returns redesigned work around AI. Departments that judge AI by at least one growth measure, such as net-new revenue or sales conversion, reported strongly positive returns at 35%, against 13% for those measuring efficiency alone. High-return departments were more likely to have changed their management structure (64% vs. 47%) and significantly changed workflows (84% vs. 69%). They did not report more staff cuts or hiring freezes.

Korst, a former Microsoft and T-Mobile executive, compared the moment to factory electrification. Factories first swapped their central steam plant for a single electric motor — cleaner, more efficient, small gains. "It took someone like Ford to come along to say, 'Well, let's rethink this,'" he said. Machines could then go anywhere on the floor, and production was rebuilt around them.

That rebuild took decades. Tambe noted electricity took roughly a century to spread through the economy, and about 50 years to reach 40% of U.S. households. "It's a great comparison in terms of the scope and the scale, but the speed of what's happening … is really something quite different," he said, noting ChatGPT accounts were everywhere within a year or two. Much of the factory reorganization came when a new generation of managers replaced the old one, he said — "So we don't have that kind of timescale."

The survey shows the rebuild has started: 52% of managers and above said their department changed its management structure because of AI, and 73% said workflows changed significantly. Tambe also cited Stanford economist Robert Hall's finding that big economic shifts tend to lock in during recessions, an argument Goldman Sachs economists David Mericle and Pierfrancesco Mei echoed last October: "History also suggests that the full consequences of AI for the labor market might not become apparent until a recession hits."

"I'm not going to predict recessions," Tambe said. What he sees is that "the workflow transformation process has kicked off. It doesn't sound like people are waiting because this technology I think is viewed as so consequential."

The open question is who judges AI's output. "It might be the case that you have generalists that run the workflow and experts who run the verification, or you might have one person who does it all," Tambe said. "That's an interesting question."