Workplace AI use nearly doubles since 2024, but only one in three employees uses it daily, according to new Marsh study
NEW YORK, November 18, 2026 — A new study from Marsh (NYSE:MRSH), the leading professional services firm in risk, strategy, people and investments, reveals a critical disconnect between rapid AI deployment and organizational transformation. Marsh’s 2026 HR Technology's Impact on the Workforce report, which surveyed more than 6,700 employees across 10 countries, finds that while 71% of workers have used AI in their jobs—nearly double the 39% recorded in 2024—only 37% use it regularly in their day-to-day work.
“The 71% figure represents one of the steepest two-year climbs in adoption that workplace technology research has ever recorded,” said Ravin Jesuthasan, Global Workforce Leader for Marsh. “However, the headline masks a stalled transformation: without adequate support, upskilling, and work redesign, having access to AI tools does not translate into effective, sustained use.”
The AI adoption gap
The report shows that AI adoption varies by role and sector. Employees in the high-tech and IT sectors use AI regularly at 59%—nearly three times the rate of healthcare (22%) and government employees (19%). Executives use AI regularly at rates significantly higher than entry-level professionals: 64% versus 20%.
This disparity reflects each cohort’s access to AI training and support. Seventy-eight percent of executives report their employers offered them AI training in the past year, compared to 44% of experienced professionals, 32% of entry-level professionals, and 20% of hourly workers. Executives have better access to mentorship, coaching, and sandbox practice environments—conditions many workers lack. As a result, executive (89%) confidence in AI's benefits now outpaces that of entry-level (60%) and experienced professionals (69%).
Paradoxically, employees in high-tech and IT, despite having the most access to AI, report the highest rate of perceived negative job impact at 26%—11 points above the global average. Their top concerns: AI increases their workload (41%), takes more time to complete tasks (40%), and creates accuracy and privacy worries (35%).
Shadow AI, more than a compliance issue
The report identifies "shadow AI," or use of unsanctioned AI tools by employees, as a growing issue. More than one in five employees report using AI tools not supplied by their employer. Noncompliance with organizations’ AI guidelines has nearly tripled among employees in the past two years, and over one in ten employees report their organization provides no AI guidelines at all.
“Shadow AI should not be viewed solely as a governance or compliance issue. Instead, it signals that capable employees are identifying gaps between organizational demands and available tools, then solving for those gaps independently,” said Mr. Jesuthasan.
Among employees who strongly agree that they have adequate digital tools, 84% use company-supplied AI and 85% follow AI guidelines. Among those who strongly disagree with the statements, those figures drop to 42% and 54%, respectively.
The communication vacuum
While many employers are issuing AI mandates, one in three employees report that no one has discussed AI’s impact on their role or department. Among those who have heard something, only 21% cite their direct manager or supervisor as the source. This absence leaves employees to interpret mandates, model behavior on what they observe, and make judgments in a vacuum about what is expected and permitted.
A lack of communication may also dissuade employees from experimenting with new technologies. Nearly one in three employees describe themselves as somewhat or very reluctant to try new technologies in the workplace.
The skills problem
Most organizations focus AI training on technical skills. But the key skills that determine if AI usage translates into performance are typically: judgment, critical thinking and contextual knowledge.
Because many executives have had more time and support to build these pivotal skills, 9 out of 10 are confident in their ability to do their jobs well when using AI tools. Further down the organization, that investment has not yet been made, and as a result, only five in ten hourly workers feel the same confidence.
Furthermore, only 30% of employees strongly agree it's easy to understand what skills they need to advance their careers.
"The practical question for leaders now is whether the conditions for responsible AI adoption are in place in their organizations. What leaders do—how they model AI use, set norms and communicate them, and invest in human skills—will determine the return on their AI investments,” said Mr. Jesuthasan.