The Rise of AI Theater
Decades ago, sociologist Erving Goffman observed that social life has the structure of theater. Whenever we have an audience, we read the situation, understand the role it calls for, and manage the impression we make. His insight was simple: people do not just perform tasks; they also perform identities. We present versions of ourselves that fit what the situation rewards.
That insight is playing out in AI rollouts today, and it’s creating one of the more unexpected AI adoption challenges leaders face. As many as one in six employees admit they sometimes pretend to use AI at work. At the same time, Microsoft and LinkedIn found that more than half of employees who do use AI hesitate to admit relying on it for their most important work because they worry it makes them look replaceable. Whether people are overstating their use or hiding it, the behavior points to the same underlying reality: AI has become more than a tool. It has become a signal. In workplaces where AI use is increasingly associated with competence, innovation, and adaptability, people begin performing the version of themselves they believe the environment rewards.
We call it AI theater. And it sharpens the question leaders keep asking us: why should AI be treated any differently from every other technology we have rolled out? Because AI asks people to do more than learn a new tool. It asks them to rethink how they create value.
AI Adoption Is Different from Traditional Technology Rollouts
Traditional technology rollouts tend to run on the same engine. There’s a pressure to adopt a new system on a hard deadline. Eventually, the new tool becomes the only way to get the work done. Other times, there is no formal deadline at all. The world simply changes around the technology until opting out stops being viable. As email and cellphones became widespread, the old ways of communicating became increasingly difficult to maintain. You could resist, but eventually the environment did the enforcing.
And the change, in the end, was contained. Email and cellphones did change how we work to some extent. Communication sped up, we became more reachable, and expectations shifted along with it. Replies were suddenly expected in hours instead of days, and being away from the office no longer meant being unavailable. More than anything, it changed how fast we worked, but it did not change how we thought about our work. We were still largely doing the same work, just faster.
Why This Change Feels Personal
AI is not following that pattern. We do not just operate it; we work with it. We hand it drafts, question its output, bring it into team meetings, and talk about it like a peer. For the first time, we are integrating a partner that is not human into the daily flow of work, and that changes more than the flow. It changes how we think about the work itself.
For the first time, we are integrating a partner that is not human into the daily flow of work, and that changes more than the flow. It changes how we think about the work itself.”
That is why identity is in play. Most people have built their professional worth on what they do and what they are known for. Now the role is being rewritten, and they are asked to hold the pen. The questions underneath are silent but persistent: Where do I fit? What is my value here now? And for specialists pouring their expertise into these tools: Am I training my own understudy?
For some, the conflict is one of belief. Many employees embrace AI, while others struggle to reconcile it with what they hold to be true, whether that’s the environmental cost of the technology, a conviction about the right way to work, or a deeper unease about what it means for people when a machine does the thinking.
We Have Seen This Play Before
AI is not the first initiative to create a gap between visible rollout and everyday practice. Organizations have been here before with ethics programs, safety initiatives, quality management, sustainability, and countless transformation efforts. Launching a new initiative is rarely the hard part. Embedding it into the way people actually work is.
Research on organizational transformation keeps finding the same pattern: many change programs labeled as transformations are really incremental efforts in planning, communication, and management that never fundamentally alter how the organization operates. Transformation becomes real only when new behaviors become natural, demonstrated habits of everyday work.
AI theater is that same gap playing out at the level of the individual employee. Signals like the number of licenses an organization has and AI usage logs tell us very little about whether work is actually being done differently. Adoption is not measured by visibility. It is measured by behavior.
Optimism bias quietly reinforces the distance: everyone acknowledges that AI will reshape work while assuming the real impact will fall on another role, another team, sometime later.”
This is also why communication alone cannot close the gap. Awareness is only the beginning of change. People can agree with the message intellectually and never connect with it personally. Optimism bias quietly reinforces the distance: everyone acknowledges that AI will reshape work while assuming the real impact will fall on another role, another team, sometime later. Not on them. Certainly not yet. Organizations respond by communicating harder, while everyday work stays exactly where it was.
How Psychological Safety Shapes AI Adoption
Why do people protect themselves this way? The clearest answer is psychological safety, or the lack of it. People take the risks that learning requires (asking questions, admitting confusion, reporting mistakes) only when they believe they will not be punished or embarrassed for it. Where that belief is missing, the struggles do not go away; they go underground.
The same dynamic plays out in AI adoption. When the environment rewards confidence and treats uncertainty as weakness, people do not stop learning; they conceal it. They practice in private, bury the failed experiments, and perform a fluency they have not yet earned. When people believe competence is rewarded more than learning, performance becomes safer than honesty. More training alone cannot fix it; it simply hands people better props for the theater.
The reverse is also measurable. Microsoft’s 2026 Work Trend Index reports that when managers create psychological safety around experimentation, employees show up to 20 points higher AI readiness and are 1.4 times more likely to become frequent users of AI. Safety is what turns quiet strugglers into open learners.
How Leaders Can Build Successful AI Adoption
People stop hiding the learning process when the conditions rewarding certainty are removed. That is design work, and it belongs to leaders. These four shifts matter the most.
1
Make it personal and concrete: In a recent field experiment across 515 companies, simply helping firms discover where AI fits their real work led them to find 44% more use cases and generate nearly twice the revenue of their peers. The lesson? Show people where AI fits into the work they already do instead of asking them to imagine abstract future use cases. Identify the high-value moments in their workflow, give them low-stakes ways to practice in those moments, and make clear what a good first use looks like. A mistake in practice should be a learning moment, not a mark against them.
2
Make struggle discussable: Equip managers to carry the conversation in their own team meetings: what peers tried, what worked, and what did not. If honest struggle reads as breaking character, learning moves out of sight, and the whole team hits the same wall separately. Psychological safety is the daily experience of raising a hand without paying for it.
3
Treat it as a culture shift, not a launch: A campaign can launch in weeks; a change of this magnitude plays out over 12 to 18 months of continuous reinforcement. People’s belief systems have to shift, and that happens only when new behaviors are seen, repeated, and reinforced daily. Our BRAVE framework names the five conditions that make this possible: belonging, relevance, access, visibility, and empowerment. It starts with belonging for a reason; that is where safety lives. When any condition is missing, you get ceremony instead of adoption.
4
Measure practice, not ceremony: Deployment, completions, and login rates count the ceremony, and ceremony is easy to manufacture; theaters even have a name for filling seats with free tickets so a show looks sold out: papering the house. The better questions: Is our workforce getting more comfortable? Are people experimenting openly? Are experiments leading to better output? Is the business actually getting better at the work?
A Human Transition, Not a Technical One
For the first time, technology is not just something we use to do the work. It participates in the work. That is reshaping the relationship between people and their work, and a relationship cannot be scripted. It is built through trust, safety, and consistent signals, reinforced over time.
If you’ll allow me to continue to extend metaphors, actors also have a phrase for the moment a script is no longer needed: going off book. The lines are not being recited anymore; they have been internalized. That is the real goal here, and it is why AI adoption cannot be run like a rollout. You can mandate a performance. You cannot mandate what people internalize.
My colleagues and I explored this question in our webinar, From AI Theater to Real Adoption.
About the Author
