Employees may resist a new tool even when it promises to save time because they are responding to what it could do to their workload, control, skills and standing at work. A 2025 review of 63 studies argues that resistance is often a response to perceived threats, not simply a refusal to learn unfamiliar software.
The review, by Veronika Cieslak and Carmen Valor of Universidad Pontificia Comillas in Madrid, examines employee resistance to digital transformation. It puts a human explanation behind a familiar workplace problem: a company buys a system to speed up routine work, but some people keep using the old process, avoid the new tool or find ways around it.
Managers may see a time-saving application. An employee may see a changing job, a new source of monitoring or a period in which hard-earned expertise is less visible. Both views can exist at the same time.
Technology can threaten more than a task
Many adoption models focus on usefulness and ease of use. Those questions still count. A system that is confusing, unreliable or badly supported will struggle to gain acceptance.
Cieslak and Valor argue that the calculation is broader. Workers can see digital technology as competing for resources they value, including employment prospects, professional identity, autonomy and relationships with colleagues.
The authors describe four pathways that may lead to resistance: burdening, diminishing, disempowering and isolating. They are a framework drawn from earlier studies, not a checklist that predicts how any individual employee will respond.
Burdening occurs when a tool adds work before it removes any. A team may have to learn a new platform while maintaining existing targets, enter the same information into old and new systems during a transition, or spend time fixing problems that did not previously exist. A future productivity gain can feel remote when today’s workload has grown.
Diminishing describes a fear that a person’s skill, role or status will lose value. An experienced employee who has built a career around a difficult process may be uneasy when software begins to handle part of it. The concern is not necessarily that the employee cannot use the new system. It may be that the system changes what their experience is worth.
Disempowering concerns control. Software can allocate work, track activity, recommend decisions or set targets. Employees may want to know what data it uses, who can change its rules and what happens when its recommendation is wrong.
Isolating refers to the social side of a job. Technology can reduce informal contact, alter routines and leave people feeling less connected to colleagues. Work provides more than a series of tasks, and a system that changes how people work together can affect that wider experience.
AI gives old concerns a new urgency
AI is not the first workplace technology to provoke anxiety. But it reaches into writing, analysis, scheduling, customer support and managerial decisions, so it can affect a wider range of office and service roles than earlier tools.
The OECD reports that skill shortages remain a barrier to adoption. Around 40% of employers in manufacturing and finance that had not adopted AI said a lack of skills was the main reason. More than half of small and medium-sized enterprises that were not using generative AI gave the same answer.
At the same time, the OECD says most workers do not need to become AI engineers. Fewer than 1% are expected to require advanced AI-specific skills such as model development or programming. For many people, the immediate need is to use, analyze and interpret data, and to know when an AI result needs human checking.
Training can reduce the fear of being unable to keep up, but it cannot answer every concern. An employee may become capable with a tool and still ask whether it will monitor them, change promotion opportunities or give an opaque system too much influence over their work.
Our report on German staff using AI before employers formalize its use shows another version of the problem. Employees may begin experimenting with public tools before an organization has offered approved systems, training or clear rules. In that setting, resistance and informal adoption can appear in the same workplace.
Consultation can expose problems before a rollout fails
Algorithmic management is software used to automate or support parts of a manager’s job, such as assigning work, monitoring activity or evaluating performance. The OECD’s employer survey found that managers often thought these systems improved the quality of their decisions. It also found widespread concerns about who is accountable for a wrong decision, how recommendations are reached and whether workers are adequately protected.
Worker consultation is not a promise that every technology proposal will be accepted. It gives people who understand the daily process a way to identify weak points before the system is imposed on them.
That can uncover issues a project plan has missed. A worker may know that a customer request often falls outside the standard process, that a metric can be gamed, or that the data fed into a recommendation is incomplete. Those observations may point to a technical flaw, a poor workflow or a real risk to privacy and fairness.
The OECD found that almost two-thirds of firms using algorithmic management reported worker consultation as one of their governance measures. A separate OECD laboratory experiment in three German manufacturing firms found that consultations among workers, managers and works council representatives could produce designs participants saw as preserving productivity gains while improving job quality. The experiment does not establish that the same result will follow in every workplace, but it gives employers a reason to treat consultation as part of system design rather than a final communication exercise.
What a better rollout asks of managers
A company can improve the odds of adoption by explaining the problem a tool is meant to solve, the decisions it will and will not make, the data it will use, and how employees can challenge an error. Training should give people time to practice before performance is judged through the new system.
Managers also need to be clear about what will change in a role. Vague assurances that technology will make work easier can create more suspicion if employees are being asked to surrender discretion or if the business has not decided how staffing will change.
Resistance does not automatically mean a technology is badly designed, and it does not give workers a veto over every investment. It can still be evidence that a rollout has left unanswered questions about workload, skills, data and accountability.
The Cieslak and Valor review is available in Cogent Business & Management. The OECD’s AI and skills report and its algorithmic management survey provide the workforce evidence used here.
