Businesses have AI tools. Why aren't employees using them?
As enterprise AI deployments accelerate, businesses find workflow redesign, change management and employee trust, not just the tools, determine adoption success.
Many businesses have moved beyond basic AI experimentation, investing in enterprise AI assistants, copilots and other generative AI tools. But business leaders have found it challenging to get employees on board with AI.
Despite significant investment, businesses continue to struggle with internal AI adoption. Deloitte's 2026 State of AI in the Enterprise report found that only 25% of organizations have moved 40% or more of their AI initiatives into production, underscoring the difficulty many businesses face in scaling AI across the enterprise. For many organizations, moving AI into production is only the beginning.
"Deployment is not the finish line. It is the starting line," said Snorre Kjesbu, senior vice president and general manager of collaboration at Cisco. "The organizations that treat it otherwise are the ones wondering why nothing is moving."
Businesses often underestimate the scale of organizational change required to successfully adopt AI, Kjesbu said. The first challenge is redesigning the organization itself. "The structure, the roles, the decision-making velocity, all of it needs to evolve," he said. "That is not a small ask, and most organizations are not moving fast enough on it."
Businesses that successfully drive AI adoption combine strong executive leadership with bottom-up momentum across the workforce, Kjesbu said. While executives need to establish a clear vision for how AI fits into the business, employees also need opportunities to explore how the technology can improve their work. That combination, he said, helps AI adoption spread organically rather than being viewed as just another top-down technology initiative.
Businesses seeing stronger adoption rates are investing in workflow redesign and governance that give employees the confidence to use AI effectively.
Workflow redesign and positive sentiment drive adoption
The reasons employees hesitate to embrace AI vary from business to business, but a common theme is the difficulty integrating new tools into existing workflows. Poorly designed AI tools, cumbersome user experiences and unclear use cases can drive employees to refuse AI altogether or seek out external AI tools instead of using those approved by their organizations.
Employees are far more likely to adopt AI when it solves an existing business problem than when they're simply asked to use a new tool, said Lian Jye Su, chief analyst at Omdia, a division of Informa TechTarget. Businesses can address adoption challenges by building a strong data foundation, redesigning workflows around AI capabilities, and promoting change management, employee training and continuous feedback to improve adoption over time.
Employee resistance to AI is more of a reflection of the organization than the employee.
Lian Jye SuChief analyst, Omdia
Workflow redesign should make it easier to use approved AI tools instead of looking for alternatives, said Mike Levin, general counsel and chief information security officer at Solera Health.
"Make the approved path faster than the workaround, train before the incident instead of after, and start from problems employees already have," he explained. "Nobody adopts a tool; they adopt a solution to something that was consuming their time in ways they did not want."
Employee attitudes toward AI also play a bigger role in adoption than many businesses realize, said Rob O'Donohue, vice president analyst at Gartner. AI use alone doesn't drive productivity -- employees with positive sentiment toward AI are more likely to be highly productive than those who just use the technology out of obligation, he said.
Su cautioned that business leaders should consider the organization's deficits when sentiment toward AI is low. "Employee resistance to AI is more of a reflection of the organization than the employee," he said, pointing to organizational and leadership factors as critical influences on adoption.
Therefore, business leaders must gauge employee sentiment toward AI and the resulting productivity to achieve better adoption rates. And rather than creating new metrics to measure internal AI adoption, executives should focus on the performance metrics employees already know and use, according to O'Donohue.
These factors are just as important as the AI model's performance in determining whether AI successfully moves beyond deployment and into employees' day-to-day workflows.
Governance should build trust, not create barriers
While investing in change management can help bolster employee sentiment, adoption and productivity, an underlying fear or distrust of AI can persist across employees.
"The struggle is multi-fold," Su said. "On the surface, there is lack of buy-in from internal stakeholders. Employees may hesitate to use AI to perform their daily tasks. When digging deeper, the reluctance often comes from a lack of trust."
Employees are often uncomfortable relying on AI for decision-making because the systems are viewed as "error-prone, probabilistic black boxes," Su added.
Most employees don't distrust AI because they're stubborn; they distrust it because they see it as a trap.
Mike LevinGeneral counsel and CISO, Solera Health
Governance isn't just about reducing risk; it can also help employees develop trust and confidence in AI. It should establish clear boundaries around how AI can and should be used, rather than becoming a barrier to adoption, Solera's Levin said. Clear policies on data, decision-making and automation can build trust by helping employees understand where AI is appropriate in their daily work and where human oversight remains necessary.
Governance should promote and protect easy-to-use AI as much as it establishes clear boundaries for it. Governance focused primarily on restricting AI use can backfire, Levin said. When approved tools become difficult to use, employees might turn to unauthorized AI applications outside company oversight, a practice commonly known as shadow AI.
"Shadow AI is the tax you pay for governance that says no by default," he said. "The programs that work make the sanctioned path the easiest path. Make the approved path faster than the workaround, train before the incident instead of after and start from problems employees already have. Nobody adopts a tool; they adopt a solution to something that was consuming their time in ways they did not want."
Solera intentionally avoids assigning AI governance to a single department, Levin said. Instead, representatives from legal, cybersecurity and business units work together from the outset, helping ensure that decisions reflect both operational and compliance considerations. He contrasted this approach with what he described as a serial review process, in which departments evaluate AI independently, with no single group taking ownership of the overall outcome.
"Most employees don't distrust AI because they're stubborn; they distrust it because they see it as a trap," Levin said. "When you put it in writing, here's what data you can use, here's where a human stays in the decision, here's what we will never automate, you encourage adoption. Clear boundaries give permission to move fast inside those boundaries."
Liz Hughes is an award-winning editor and writer covering AI and emerging technology and the former editor of AI Business and IoT World Today.
