How employee involvement helps turn AI investments into results

AI results depend on changing how people work

Only one in five CEOs say their companies are achieving most or all of the AI results they intended. Transformation advisers with 40 years of experience helping companies through technology changes report this finding from their latest CEO survey. They identify three recurring obstacles: employees lack the skills and tools they need, organizations remain focused on separate pilots, and people hesitate to use AI across their work. The advisers benefit when companies invest in managing these changes.

The advisers say these human obstacles are worsening each quarter, making adoption a growing competitive concern. As AI changes industry economics, companies learn where it can improve their position and adjust how they work. Employees contribute knowledge that makes those practices specific to the business. Continued learning builds an advantage over time, so delaying adoption also delays the experience a company can gain.

Yet employees are being asked to join changes whose effects on their working lives remain uncertain. AI raises questions about employment, career prospects, professional identity, and continued relevance. Managers and frontline employees determine whether an investment becomes part of everyday work, so unanswered concerns can block the results leaders expect. Leaders need to explain why participation is worthwhile for the people whose work will change.

Cost-led AI strategies give employees reason to resist

Employees have a clear reason to hesitate when leaders explain AI mainly through lower labor costs. The advisers describe the preceding three years as dominated by messages about efficiency and productivity inside organizations. Productivity is a legitimate business objective, but employees who hear that fewer people will be needed have reason to question what successful adoption means for their careers. More enthusiastic communication cannot remove that conflict.

Klarna made the conflict visible by presenting AI as a way to replace employees. Its workforce shrank 40% between December 2022 and December 2024. In May 2025, its CEO publicly acknowledged that the company had placed too much emphasis on efficiency and cost and said it was actively rehiring. Employees therefore had to weigh the company’s invitation to adopt AI against its stated intention to reduce staffing.

Duolingo faced a related communication problem when a message about staffing and automation drew a negative response. Within weeks, its CEO acknowledged handling the communication poorly. Both companies left employees considering how changes to their work might reduce their future opportunities. A message centered on savings can therefore weaken the willingness to participate that implementation requires.

Concern about future opportunities also appears in polling of Gen Z workers. Findings from Gallup’s Voices of Gen Z, conducted February 24 to March 4, 2026, with 1,572 respondents, and the Quinnipiac University Poll, The Age of Artificial Intelligence, from March 2026, describe both trust in AI-assisted work and expectations about jobs.

These concerns become harder to address when managers and frontline employees are excluded from decisions about how they will work. Middle managers matter especially because they can be anxious about the change while also being responsible for implementing it. Microsoft’s CEO has described a future in which people supervise much greater machine capability; Microsoft sells AI tools and benefits from wider adoption. Employees need that future translated into practical decisions about what they will do, which judgments they will make, and how they can influence the work being redesigned.

Give employees a practical reason to use AI

A practical explanation starts with improving work employees already understand. When AI reduces cumbersome tasks or unnecessary transfers of work between people, employees can judge the benefit in their own day. The Adecco Group, Presidio, and an unnamed consumer packaged goods company approached that opportunity in different ways. Their reported experiences show how the choice of problem can shape employees’ response to a change.

At employment agency The Adecco Group, the first participants were employees already interested in learning about AI. These willing learners helped explore applications, giving colleagues experience they could observe before deciding to participate. Recruiters identified tasks that were cumbersome or added little value from a human perspective. AI reduced those tasks so they could spend more time building relationships with people.

By choosing the tasks, recruiters connected the technology to work they valued. Denis Machuel, CEO of The Adecco Group, put the principle directly: “AI has to happen with people, not to people.” Adecco benefits commercially when its recruiters work more productively and serve people effectively, so Machuel has a stake in this approach succeeding. The company reported that recruiters became more productive and enjoyed their work more as low-value tasks took up less of their day.

Their experience gave colleagues a reason to try AI themselves. “When you start to do that,” Machuel said, “then the word of mouth goes across the whole enterprise, like, ‘well, I enjoy my work more because I‘m doing all the stuff that I love,’ [and] then adoption scales.” His account describes adoption spreading as employees told others about work they enjoyed more. Colleagues could understand the benefit because it came from people doing familiar jobs.

Presidio chose a problem employees already wanted solved. Paula Cipollone, Chief Transformation Officer at the digital solutions and services provider, described announcing that the team would repair a process employees had disliked for years. The room responded with a standing ovation. Presidio sells services that help organizations change how they use technology, so successful transformation supports both its internal operations and its commercial offering.

The enthusiasm Cipollone described came from a promise to fix a recognized problem. Choosing that problem also gave employees a clear way to assess the eventual change: they could compare how the process worked before and after the repair. Adecco’s account went further, describing employees using AI and sharing their experience. Both examples begin with a difficulty employees could identify, but the reported outcomes occur at different stages of implementation.

An unnamed Fortune 500 consumer packaged goods company addressed daily work by changing how it explained the transformation. Its chief digital transformation officer described moving away from an emphasis on savings and adopting the idea that “simplifying work plus digital equals easy.” The company reported greater momentum after focusing on making employees’ work easier. Savings could remain a business objective while the reason offered to employees addressed their daily experience.

Connect better work to a direction employees can help shape

Improvements to daily work give leaders a starting point for describing a broader working future. Repetitive activities can take less time, leaving employees more attention for judgment, relationships, and other responsibilities they value. To make that direction practical, leaders need to redesign workflows alongside skills development and employee involvement. Employees can then understand how their role changes and help determine whether the new process works.

That direction also needs an honest explanation of business urgency. The advisers argue that competitors learn, innovate, and gain market share while employees help build knowledge and practices specific to their own company. Leaders should explain how delayed learning affects that position and what participation can make possible for each person. Personal benefit and competitive urgency can support the same effort without promising permanent job security.

Purpose can give employees another reason to contribute beyond an easier working day. Glenn Fogel, CEO of Booking Holdings, whose businesses include Booking.com, Priceline, OpenTable, Kayak, and other travel sites, emphasizes shared achievement. As the company’s leader, he benefits when that motivation improves team performance. “Some [companies] are more successful than others,” he said. “And part of the reason is because the people on the team … have an objective they want to win. They’re not showing up just to get a paycheck. They’re there to accomplish a mission, to achieve something.”

A mission becomes credible when employees can see progress and influence what comes next. Drawing on their decades of transformation work, the advisers identify clear direction and supportive employee networks as useful practices. With AI, that direction needs room for revision as technology develops and the company learns. Sharing early successes, recognizing contributions, and inviting feedback give employees a way to help shape the next stage.

Give employees responsibility for redesigning their work

Helping shape the next stage requires a say in decisions about how work will change. An unnamed transformation officer at a consumer packaged goods company reported that middle managers find ways around working practices imposed on them. Those managers understand the routines and constraints that determine whether a process can work. Giving them design responsibility brings that knowledge into the decision before they respond by finding workarounds.

Design responsibility also makes accountability more workable. A manager asked to explain a completed decision has little influence over whether it fits the team’s work. A manager involved in developing the process can identify difficulties, test changes, and revise them with colleagues. Responsibility for the outcome then comes with a chance to improve the decisions that produce it.

Testing changes requires permission to encounter difficulty. The advisers describe AI technology as changing weekly, making immediate mastery an unrealistic expectation for managers and frontline teams. Leaders should explain that early mistakes are part of learning and provide tools for testing useful applications. Without that permission and support, uncertainty can lead employees to wait for someone else to establish a safe approach.

Peer champions can help employees begin those tests and learn from them. At Presidio, champions within individual functions received central training, then shared useful practices and collected feedback from colleagues. This combined common preparation with knowledge of each function’s work. An unnamed Presidio client asked to reproduce the program, showing interest in the approach beyond the company itself.

The champions’ feedback role lets learning travel in both directions. Employees can ask colleagues about a task, discuss a useful practice, or work through an unsuccessful attempt. Champions can then report recurring difficulties to the people coordinating the transformation. Decisions made centrally gain information from actual use, allowing the program to change as employees discover what works.

Independent experimenters offer another way to gather that knowledge. Some employees already try AI in their own time, finding applications and developing judgment about where it helps. Leaders should identify them, provide appropriate tools, and listen to their discoveries. Bringing those discoveries into formal programs gives the programs access to experience beyond the applications chosen centrally.

Employee influence requires continued executive support. Managers need permission to experiment, champions need training, and independent discoveries need a route into organizational decisions. Leaders must support those arrangements and respond to what employees learn. Their own behavior helps employees judge whether that support will last.

Leaders must demonstrate the change they expect

Employees can first judge executive commitment through how leaders spend their time. Christophe de Vusser, Bain’s CEO and Worldwide Managing Partner, devotes 20% of his time to the firm’s AI transformation. Bain sells transformation advice, so successful AI adoption supports its own operations and its standing with clients. De Vusser’s time commitment makes the work a continuing executive responsibility that employees and clients can observe.

Direct use gives employees another way to assess that commitment. Bob Bradway, CEO of Amgen, put the entire senior executive team through AI courses and uses AI himself. His experimentation includes vibe coding, in which a person uses natural-language instructions to have AI generate and revise code. He has also used AI to work through questions over a weekend that he would otherwise have assigned to his team.

Bradway has a stake in successful adoption because Amgen benefits when its executives and employees use AI productively. He connects his own participation to employees’ willingness to engage: “If the CEO is enthusiastic about AI, after a while, people will start to figure out whether they’re enthusiastic, too,” he said. His behavior makes learning and experimentation visible at the top of the organization. The advisers express this leadership principle as “show, don’t tell.”

Microsoft’s research connects everyday team and manager behavior with AI use and trust. The company has a commercial interest in these findings because it sells AI tools and benefits when organizations adopt them. Its findings include trust in agentic AI, meaning systems that can take steps toward a goal with some autonomy. They also describe collaboration among users and employees’ perceptions of leadership alignment.

These findings make leadership consistency a practical implementation issue. Employees need to know whether permission to experiment, investment in training, and expectations for adoption have support across the organization. A CEO’s enthusiasm cannot establish that consistency alone. Leaders at each level need to demonstrate use, support learning, and respond to their teams’ feedback.

Address career concerns as the work changes

Consistent support still needs to address what AI means for employees’ futures. An easier working day can coexist with concern about employment, identity, and professional relevance, as Klarna’s replacement message illustrates. Employees may find good reasons to use AI while remaining unsure about their career prospects. Leaders need to discuss those concerns directly rather than assume participation has settled them.

Those conversations must continue because AI capabilities keep developing. Employees will face further questions about which responsibilities they retain and how their expertise remains valuable. The advisers’ familiar transformation practices, clear direction, involvement, and support, remain useful, but working arrangements will continue to change. Executive responsibility therefore extends beyond the point when employees begin using the technology.

An unnamed Fortune 500 company faced that responsibility as it approached go-live, the point when a new system or process enters everyday use. Teams began saying they were unready, concerns escalated, and resistance grew as trust deteriorated. Announcements failed to resolve the implementation problem. The chief transformation officer convened in-person sessions where employees could discuss their worries and work through feasible solutions.

Those sessions made readiness something employees and leaders could examine together. By being present, the officer could hear concerns directly and discuss responses with the people expected to use the changed process. The company reported renewed progress after the conversations. Employees’ resistance had given leadership information about implementation problems that required direct attention.

The bottom line

AI results depend on whether employees can use the technology to improve work and help decide how that work changes. Leaders who frame adoption mainly around labor savings give employees reason to hesitate. A more credible approach connects business goals with practical improvements employees can recognize, while addressing career concerns without promising outcomes the company cannot guarantee.

For executives, this means treating employee involvement as part of implementation. Managers need responsibility for redesigning workflows, teams need tools and time to learn, and employee feedback needs a route into decisions. Leaders must also demonstrate the behavior they expect by using AI themselves and supporting experimentation consistently.

Adoption is not complete when a tool goes live. As capabilities develop, leaders will need to revisit roles, skills, and working arrangements with the people affected. The experience employees gain through that process helps the company learn where AI delivers value. Delaying their involvement delays that learning, even when the technology is already available.