"Employees will judge an AI announcement by the incentives and decisions behind it, not by a polished slogan."

About Dr. Gleb Tsipursky

Dr. Gleb Tsipursky is a behavioral scientist and CEO of Disaster Avoidance Experts, where he advises organizations on practical AI adoption. He is the author of The Psychology of AI Adoption at Work: From Resistance to Results (Georgetown University Press, 2026), with a foreword by Stanford economist Nick Bloom.

When an AI rollout stalls, how often is the technology actually the problem? What is usually going on instead?

There is no reliable universal percentage, and I would distrust anyone who gives you one. Sometimes the tool genuinely falls short: it lacks access to the right information, produces errors, or does not fit the workflow. But executives often discover that the software performs adequately in a demonstration while employees avoid it, work around it, or spend so much time checking outputs that the promised efficiency disappears.

The missing ingredient is usually attention to the people who must use the system. They may fear mistakes, worry about their jobs, lack time to practice, or see little value in changing a process they know well. Leaders should examine the actual task, the output quality, and the human workflow together. Ask employees to demonstrate where the system helps and where it adds rework. Track adoption alongside time saved, accuracy, and staff confidence. That reveals whether to fix the technology, redesign the handoff, or change how people are trained.

You connect “automation anxiety” to loss aversion, and note that experienced employees feel it most because AI can threaten an identity built over a whole career. How should a CEO talk to a 20-year veteran who suspects AI makes their expertise worthless?

I would begin by acknowledging what the employee has spent two decades building. Experience is an asset precisely because it includes judgment about unusual cases, customer needs, and warning signs that rarely appear in a procedure manual. When a leader introduces AI by praising its speed alone, an experienced worker may reasonably hear that their hard-earned competence is being discounted. Loss aversion magnifies that threat.

I would say, “Your experience matters in deciding which parts of this work can be assisted, what the system gets wrong, and what standards we keep.” Then I would make that statement operational. Invite the employee to help choose one real workflow, identify examples that challenge the system, and establish a review step. Give them time to learn, and recognize the contribution publicly. Some tasks will change or disappear, and pretending otherwise damages trust. The goal is to give an experienced person a visible role in shaping the change rather than treating them as an obstacle to it.

One of your middle-manager profiles, the Wait-and-See Traditionalist, keeps asking for one more pilot before committing. How can an executive tell honest caution from quiet resistance, and what actually moves that manager?

I would avoid deciding what kind of person the manager is based on a label. Start by asking for the specific evidence that would make them comfortable moving forward. Honest caution usually names observable concerns: inaccurate answers, sensitive data, poor integration, customer risk, or inadequate training. Those concerns can be tested. Quiet resistance often shows up when the conditions for approval keep changing, even after the evidence requested has been supplied.

A CEO can turn an open-ended debate into an explicit decision process. Choose a bounded workflow, agree on quality and time metrics, specify the cases requiring human review, and set a date to evaluate the result. Ask the manager to help design the test and own the decision criteria. If the pilot misses those criteria, fix the issue or stop it. If it meets them, move to a supervised rollout instead of commissioning an indefinite series of new pilots. Accountability should run both ways.

Many leaders introduce AI as an efficiency drive, and employees hear “layoffs.” What should a leader say, and commit to, in that first announcement? And what should they never say?

Employees will judge an AI announcement by the incentives and decisions behind it, not by a polished slogan. Leaders should describe the actual business objective, the workflows being considered, how the organization will measure quality and time savings, and what will happen to capacity that is freed up. They should explain who can challenge errors and how employees will participate in choosing and testing tools. If the company intends to use gains for better customer service, reduced backlogs, or growth, say so and show the plan.

If staffing reductions are possible, do not promise that no jobs will change. People will remember that broken promise long after the rollout. A credible commitment is to communicate decisions plainly, provide training and transition support, and avoid evaluating employees on tool use alone before they have had a fair chance to learn. Never say, “AI is coming whether you like it or not.” It converts uncertainty into opposition.

If an executive reading this could do one thing in the next 30 days to move their organization from resistance to results, what would it be?

Choose one recurring workflow with a clear owner, involve the people who perform it, and run a small, measurable experiment. Avoid starting with a companywide mandate or a list of tools. First, record the current time per task, the typical quality problems, and how often corrections are needed. Have the team identify where AI might help and where a human must make the final decision. Use a modest set of real, authorized examples and agree on how to check the results. Then let employees test the workflow during paid work time, not as a homework assignment.

Review the evidence together after two to four weeks. Did it save time after checking and rework? Did errors fall? Are people confident using it? What did they refuse to trust, and why? Keep what works, repair what does not, and give the team a voice in the next experiment. That is how adoption becomes observable work rather than a slide in a strategy deck.

Latest From The Executive Insight