From operational discipline to organizational friction
As organizations push further into agent-based models, employee resistance has increased to 20%, up from 5% last quarter. The nature of that resistance, however, has shifted.
Resistance is increasingly driven by trust and ethical considerations (53%), while concerns around increased workload or complexity have nearly doubled, rising from 28% to 51%. At the same time, concerns over skills gaps have eased, dipping from 76% to 57%, alongside a sharp decline in concerns about insufficient training or support (from 53% to 27%). Concerns about job security have also moderated, falling from 67% to 55%.
At the same time, some organizations are experimenting with unconventional and often counterproductive ways to drive usage. "Token-maxxing," a practice that gamifies token consumption through incentives and leaderboards, has emerged as a controversial culture hack. While 41% of leaders say they would consider it, sentiment remains mixed with many cautioning against equating activity with impact (22% opposed and another 37% neutral on the practice).
“Token-maxxing presents a classic risk of incentivizing activity over outcomes,” said Edwige Sacco, Head of Workforce Innovation at KPMG LLP. “What starts as a productivity metric can quickly become a distraction. In the short term, it deteriorates value; in the long term, it undermines culture.”
Value over volume
In an era where operational discipline matters as much as driving usage, learning to use AI effectively takes on new importance. “The organizations best positioned to derive the most value from the tokens they use are the ones building the habits, skills and governance structures that make every interaction count,” continued Shears.
Nearly half of leaders (47%) say employees who use AI effectively are already outperforming their peers, a clear signal that the competitive advantage aligns to how well people use AI to create organizational value.
The KPMG Quarterly AI Pulse Survey in the U.S. captured perspectives between April 28th and May 25th from 204 U.S.-based C-suite and business leaders representing organizations with annual revenue of $1 billion or more. More than a third have revenues of $10 billion or more. The US insights are part of KPMG’s Global AI Pulse research.
Additional findings from the KPMG Q2 AI Quarterly Pulse Survey are below:
- Leaders plan to invest a weighted average of $202 million over the next 12 months, on par with last quarter’s $207 million.
- 35% of leaders say that AI cost management and economic literacy – including understanding usage-based pricing models such as token and inference costs – remains a barrier.
- While two-thirds of organizations have monitoring dashboards (66%) and approval processes (61%) in place, fewer have implemented direct token/usage controls (36%), indicating that true token-level management is still maturing.
- 32% say CEO or an executive committee member is accountable when making a business decision. 34% say it’s another member of the C-suite.
- Upskilling and reskilling remains the top strategy to meet the needs of an AI-enabled workforce for the second quarter (65%).
- 65% say teaching prompt and instruction skills remain the most important way to train employees to work productively with AI agents.
- When asked how much more they would pay for candidates with strong AI skills, 40% of leaders said they would offer a 6–10% premium, while 38% said they would pay 11–15% more compared with candidates without those capabilities.
- 54% agree or strongly agree that in today's labor market, strong social and interpersonal skills are more important for career success than strong mathematical or technical skills.
- 55% expect increasing levels of AI fluency within their organization and roles will change for employees who do not adopt these capabilities over time.
