Wharton's Fourth Annual AI Survey: Most Companies See Returns, Few See Big Ones

Four years into enterprise AI adoption, the interesting question isn't whether companies are using the tech anymore. It's why the returns are so lopsided.

Wharton Human-AI Research (WHAIR), working with GBK Collective, just released its fourth annual study on enterprise AI adoption, and the topline numbers tell a split story. Eighty percent of enterprise leaders report positive returns on their AI investments. Only 28% call those returns strongly positive. That's a 52-point gap between "it's working" and "it's actually moving the needle," and the report spends most of its length explaining why.

The study surveyed 1,416 U.S. enterprise leaders and entry-/mid-level employees, and it identifies three markers that separate the winners from everyone else: proving value, transforming workflows, and aligning the organization around AI. Here's the kicker: only 7% of organizations hit all three.

"Over four years, we have watched AI move from experimentation into everyday business operations. This year, the question changes," said Stefano Puntoni, Sebastian S. Kresge Professor of Marketing at Wharton and WHAIR Faculty Co-Director. "The vast majority of companies are using AI and most are seeing some return. What separates the leaders is whether they are changing the business around AI enough to generate meaningful value."

The Metrics Problem Nobody's Fixed

If you're a practitioner who's spent any time arguing for AI budget, this part will sound familiar. Eighty-three percent of leaders expect AI to fundamentally transform or majorly impact their industry within two to five years. But the most common ROI metric departments use is still hours saved, cited by 40%. Net-new revenue, arguably the metric that actually matters for growth, is the least-used major measure at just 27%.

That metric choice isn't cosmetic. It correlates directly with outcomes:

• Departments tracking at least one growth metric (net-new revenue, sales conversion, output volume, customer satisfaction) report strongly positive returns 35% of the time
• Departments measuring efficiency alone see that number drop to 13%

"Companies believe AI is going to transform their industries, but many are still asking it to prove itself by saving time," said Prasanna "Sonny" Tambe, WHAIR Faculty Co-Director. The companies pulling ahead, he notes, are redesigning workflows and putting ownership structures in place rather than just tracking time saved.

Despite the stated ambition, 52% of leaders still say AI's main job over the next 12-24 months is cutting costs or improving productivity. Only 22% point to growth: new revenue, new products, better customer experience.

Organizational structure matters too, and the numbers are stark. Among managers and above, 58% at companies with a dedicated Chief AI Officer report strongly positive ROI. That drops to 24% where AI oversight got bolted onto an existing executive's job, and craters to 8% with no formal AI leadership role at all. Standardization tells a similar story: 50% strong ROI at companies on one approved AI platform, versus 9% at companies still experimenting without a defined approach.

On the agent side, adoption is real but shallow. Sixty-seven percent of leaders report using AI agents in some form, and 73% say AI has significantly changed how their department works. But only 28% say AI is fully embedded in core business processes, and a mere 17% let an agent act without a human approving every single step.

"Most companies are sitting in a supervised middle, paying for autonomy and then reviewing every action, and that is where returns are weakest," said Brian Smith, EVP of the AI, Cloud & Enterprise Technology Practice at GBK Collective. The organizations doing better have explicitly decided which tasks an agent owns outright, added automated monitoring, and stopped requiring sign-off on everything.

Only one in five organizations clear all three workflow-transformation bars the study tracks: significantly changed workflows, AI embedded in core processes, and active agent use. But the payoff for clearing them is large. Among that group, 67% report strongly positive ROI, more than double the 28% baseline across all leaders.

Worth noting: strong-ROI companies aren't cutting headcount more aggressively than anyone else. They're more likely to have changed management structures, redesigned workflows, added monitoring, folded AI into performance reviews, and actually trained people.

Key Takeaways

The most striking new finding this year comes from surveying employees alongside leaders for the first time, and the gap is consistent across nearly every question. Seventy-three percent of leaders say workflows have significantly changed; only 58% of employees agree. Fifty-two percent of leaders say management structures have shifted because of AI, versus 36% of employees. Training shows the widest split: 78% of leaders say employees get adequate AI training, but only 56% of employees back that up.

Confidence and vision follow the same pattern. Seventy-nine percent of leaders say employees feel confident using AI daily, versus 62% of employees who say so themselves. Seventy-nine percent of leaders say leadership has articulated a clear long-term AI vision; just 64% of employees see it that clearly.

This isn't employee pushback. Seventy-three percent of entry- and mid-level employees say AI has had a positive effect on their own role, and 84% of those using it say it saves them time. The gap is about organizational follow-through, not enthusiasm.

"Access to powerful AI is quickly becoming ubiquitous," said Jeremy Korst, Partner at GBK Collective and Wharton lecturer. "The advantage comes from what an organization can do with it: redesign workflows, align leaders and employees and translate AI into measurable business outcomes."

The workforce picture is murkier still. Thirty-seven percent of leaders report AI-related staff reductions, 33% report slowed or frozen hiring, and 24% expect fewer entry-level hires over the next two to five years (versus 14% expecting fewer executive roles). Yet 82% say AI helps newer employees contribute faster, a genuine tension between making junior roles more productive and shrinking the number of junior roles available.

Skill atrophy worries are climbing too: 48% of VP+ leaders and 46% of entry-/mid-level employees worry their own cognitive skills are declining, versus 32% of managers and directors. And usage is diverging from sentiment for the first time in the study's history. Daily use among managers and above jumped roughly 16 points to 62%, with 91% now using AI more than weekly, but those reporting they feel "excited" dropped eight points year over year, while "cautious" and "skeptical" each rose seven to eight points. People are using the tools more and trusting them less, which is probably the stat worth sitting with longest.