Key Takeaways
- Daily AI use has become mainstream: 46% of business leaders now use generative AI daily, up from 11% in 2023, with 82% engaging weekly
- ROI tracking is now standard: 72% of enterprises measure structured ROI metrics, with roughly three in four reporting positive returns
- The talent gap is the #1 barrier: Recruiting AI-skilled workers (49%) and training current employees (46%) are now bigger challenges than technology itself
- Scale creates complexity: While smaller enterprises see faster returns, large organizations ($2B+ revenue) face integration challenges across sprawling operations
- A leadership disconnect exists: Senior executives are twice as likely as mid-managers to believe their organizations are adopting faster than peers, creating potential misalignment
- Budgets are growing with discipline: 88% expect budget increases, but 11% now comes from reallocated funds rather than new money, showing more strategic resource allocation
- The workforce impact remains uncertain: While 89% emphasize augmentation over replacement, 43% worry about skill decline among junior employees
_________________________________________________________________________________________________________________________________
A new Wharton study reveals that enterprises are embedding generative AI into daily operations, measuring ROI, and seeing positive returns. Yet the real constraint isn’t technology — it’s people, processes, and the growing gap between leaders and laggards.
The conversation around generative AI in the enterprise has shifted dramatically. Three years after ChatGPT’s debut sparked a wave of curiosity and cautious experimentation, AI has moved from boardroom buzzword to everyday business tool. But as adoption accelerates and budgets grow, a new reality is emerging: technology is no longer the bottleneck. People are.
The latest research from Wharton Human-AI Research and GBK Collective — the third annual study tracking enterprise AI adoption—paints a picture of maturing usage patterns, growing confidence in returns, and an inflection point where success increasingly depends on talent, training, and organizational culture rather than just technology deployment.
Daily Generative AI Usage Becomes the New Normal
The numbers tell a striking story of normalization. Nearly half (46%) of business leaders now use generative AI daily, up 17 percentage points from last year and a dramatic 35-point jump from 2023. More than eight in ten (82%) engage with AI tools at least weekly, cementing what the report calls “everyday AI.”
This isn’t superficial adoption. Familiarity and expertise have deepened across the board, with 77% of decision-makers reporting at least moderate knowledge of generative AI, and nearly one-third (32%) identifying as experts, up eight points from 2024. The steepest expertise gains came in Legal (up 23 points), Purchasing/Procurement (up 14 points), and IT (up 11 points).
But adoption remains uneven. Tech/Telecom, Professional Services, and Banking/Finance sectors lead, with 90% or more using AI at least weekly. Meanwhile, Retail and Manufacturing lag behind, a somewhat surprising finding given the multitude of potential applications in customer experience, supply chain optimization, and workforce management.
From Efficiency Tool to Strategic Asset
The most common use cases center on practical productivity gains: data analysis (73%), document summarization (70%), and document editing/writing (68%). These aren’t flashy applications—they’re the daily workflows where AI proves its value through time saved and quality improved.
More telling is how specific functions have found their fit. IT professionals index 123% higher than average on code writing and generation, HR teams index 129% higher on employee recruitment and onboarding, and Legal departments index 133% higher on contract generation. These specialized applications suggest organizations are moving beyond general-purpose experimentation to embedding AI in core functional workflows.
The performance data backs this up. Unlike last year, when usage and satisfaction weren’t always aligned, the top use cases in 2025 are also the top performers. Organizations are getting better at identifying where AI adds value and focusing their efforts accordingly.
The ROI Story: Positive Returns, With Complexity for Large Enterprises
Perhaps most significant for the enterprise case: accountability has arrived. Nearly three-quarters (72%) of business leaders now track structured ROI metrics tied to profitability, throughput, and workforce productivity. This represents a fundamental shift from pilot-stage enthusiasm to performance-based justification.
The returns appear real. Roughly three in four enterprises report positive ROI, with four in five expecting positive returns within two to three years. Conviction is building: 88% anticipate budget increases in the next 12 months, with 62% projecting growth exceeding 10%.
However, scale introduces complexity. While smaller Tier 2 and Tier 3 enterprises report faster ROI realization, larger Tier 1 organizations ($2B+ revenue) are more likely to report “too early to tell” outcomes as they navigate integration challenges across sprawling operations. Industry patterns mirror this dynamic: digitally-native sectors like Tech/Telecom (88% positive ROI) and Banking/Finance (83%) outpace Manufacturing (75%) and Retail (54%), where physical operations complicate implementation.
Budget allocation tells another story about maturation. About 30% of technology budgets now flow to internal R&D, signaling that enterprises aren’t content with off-the-shelf solutions. They’re building custom capabilities. And while most AI investment still comes from new budget, a growing share (11%, up seven points) comes from reallocation, often from legacy IT and HR programs, evidence of increasing discipline around where to place bets.
The Human Capital Constraint: Where Success Stalls
As the technology proves itself, human factors emerge as the defining constraint. The challenges are multi-layered and persistent.
The talent gap looms large. Recruiting workers with advanced gen AI skills ranks as the top challenge (49%), followed closely by providing effective training for current employees (46%). This comes as investment in training has actually softened (down eight points), and confidence that training alone will build fluency has dropped 14 points. Organizations are increasingly looking to hire new talent, yet recruiting remains difficult.
Leadership is consolidating but approaches vary. Executive leadership involvement has surged (67%, up 16 points from last year), and 60% of enterprises now have Chief AI Officers or similar roles. Strategy has moved decisively into the C-suite. Yet responsibility remains distributed, with 97% relying on internal teams rather than outsourcing—a sign that organizations view AI strategy as core competency, not something to delegate.
The workforce impact remains ambiguous. For three consecutive years, leaders have emphasized AI’s role as augmentation (89% agree) over replacement (71% agree). Yet new concerns are emerging. Forty-three percent of leaders worry about declines in skill proficiency, particularly among junior employees whose foundational skills may atrophy if AI becomes too much of a crutch. Senior leaders are split on whether AI will mean more or fewer hires in coming years, with junior roles facing the most uncertainty.
Culture and change management persist as barriers. Beyond technical challenges, enterprises continue wrestling with maintaining employee morale (43%) and having leadership capable of effective change management (41%). For the 16% of organizations classified as “laggards,” using AI weekly or less, employee resistance and lack of trust rank even higher as concerns.
The Seniority Divide: Optimism at the Top, Realism in the Middle
One of the more revealing findings involves the disconnect between VP-level executives and mid-managers. Senior leaders are twice as likely to believe their organizations are adopting “much faster” than peers (56% vs. 28% for managers). VPs report more positive ROI perceptions (81% vs. 69%) and greater optimism about business impact.
Mid-managers, closer to day-to-day implementation, paint a more nuanced picture. They’re more likely to emphasize employee-led approaches, reporting higher rates of investment in training programs and support for employee-driven innovation. They’re also more attuned to friction points, the gap between executive enthusiasm and operational reality.
This divide matters. If those responsible for implementation see different challenges than those setting strategy, organizations risk misallocating resources or missing the cultural shifts necessary for sustainable adoption.
What 2026 Might Bring
The report positions 2025 as a potential inflection point, the year enterprises shift from “accountable acceleration” to “performance at scale.” The infrastructure is falling into place: ROI metrics are standard, budgets are growing with discipline, and guardrails are maturing alongside broader access.
But unlocking that next phase depends on solving the human capital equation. Organizations need to close the expertise gap through some combination of training, hiring, and redesigned work processes. They need to align leadership vision with frontline reality. And they need to address the cultural elements — morale, change management, trust — that remain stubbornly persistent as barriers.
The technology is ready. The question is whether organizations can move fast enough on the people side to capitalize on it, or whether the growing divide between AI-enabled leaders and struggling laggards will widen into a more permanent competitive gap. Three years in, the AI story in the enterprise is no longer about what the tools can do. It’s about whether organizations can build the human capabilities to use them effectively.
This article was originally published on LinkedIn.
Read more of my coverage:
Mitel Introduces CX 2.0: Enterprise-Grade AI Meets Hybrid Flexibility
Cisco AI Readiness Index 2025: Why Only 13% of Companies Are Prepared for AI Success
Zoho One’s Major Update Shows How Business Software Should Actually Work