AI adoption stalls while play skills atrophy

AI adoption – A global workplace engagement drop and weak measurable returns from artificial intelligence are converging with an overlooked problem: many managers and teams are too depleted—and too deprived of play and improvisation—to handle the uncertainty AI brings.

Walk into a typical office conversation and the pattern repeats. Ask professionals what share of their day is improvised and you’ll hear answers that cluster around 60%, 70%, or 80%. Maybe more. Then ask how many have had professional improvisation training. The silence that follows does the explaining.

The mismatch isn’t just cultural. It’s strategic. As AI moves more decisions faster, the job itself can become less predictable, with higher stakes and more ambiguity. The skills the moment demands—thinking on your feet. tolerating uncertainty. responding to what’s unexpected without panicking—are the same skills many people were trained out of long before their first performance review.

Dr. Stuart Brown. founder of the National Institute for Play. has spent six decades studying what happens when humans don’t play. He defines play as “a state of mind that one has when absorbed in an activity that provides enjoyment and a suspension of sense of time.” For Brown. play is about mental attitude and motivation—not about the behavior itself.

In improvisation, that state is entered by “letting themselves say yes and seeing what happens.” It’s a deceptively simple mechanism in a world that has long treated play as secondary or suspect—reserved for actors, creative artists, and comedians, not for professional life.

That suspicion is starting to show up in the numbers, even if it’s not being fully named in the playbooks being rolled out alongside AI.

Gallup’s 2026 State of the Global Workplace report says global employee engagement has fallen for the second consecutive year to 20%. the lowest since 2020. costing roughly $10 trillion annually. The same report points to an MIT study finding that 95% of companies have seen no measurable results from their investment in artificial intelligence. In Gallup’s findings, only 12% of employees say these tools have meaningfully changed how they work.

Manager engagement adds another blunt pressure point. It has dropped nine points since 2022. and Gallup’s report says managers are now only as engaged as the people they lead. It also shows that the single strongest predictor of whether employees adopt AI is whether their direct manager champions it.

That leaves organizations trapped in a contradiction: they need managers to drive AI adoption. but those managers are among the most disengaged people in the system. They’re being asked to champion a tool that requires curiosity. flexibility. and willingness to experiment—precisely the qualities that are hardest to sustain when people are depleted. The solutions being offered tend to stay in familiar territory: more training, mandates, and adoption dashboards. What’s missing is a direct answer to what got managers depleted in the first place. and what would restore the curiosity and energy required to champion anything at all.

There’s a second failure mode running alongside depletion: deprivation. Ethan Mollick at the Wharton School. the University of Pennsylvania’s business school. has described the output of burned-out teams using AI as “workslop”—content that looks like progress. reads like progress. and quietly makes everyone a little dumber. The same phenomenon has been referred to as “brain rot.”.

In practice, the damage often doesn’t arrive as a single collapse. It accumulates through meetings and layers: meetings generate AI summaries that generate follow-up meetings. Each added layer increases cognitive load for a brain already making tens of thousands of decisions each day.

The teams that perform best, Mollick’s critique suggests in spirit if not in name, respond by subtraction. They cut meetings without clear decisions, consolidate tools instead of stacking new ones, and protect focus time as non-negotiable. They create more capacity to do good work.

But subtraction doesn’t refill the tank.

A frazzled nervous system cannot think creatively, absorb new tools, or adopt new ways of working. It cannot rest—not really. Rest is described as a regulated state you have to be able to reach. Yet people are often prescribed rest while still stuck in low-grade threat response. then return from holidays still “empty.” Play. in Brown’s framework. becomes the reset—voluntary. low-stakes. genuinely absorbing play. not a ping-pong table in the breakroom or a checkbox on a wellness calendar.

The argument is not that AI eliminates the need for humans. It’s that the more artificial intelligence is implemented in the workforce. the more human the work has to be—because AI can generate. process. and pattern-match at scale. but cannot think under genuine uncertainty. build trust between humans. or deliver the nuanced judgments that determine whether organizations survive. Those capabilities rely on a brain that is flexible. curious. and regulated—and. in Brown’s terms. a brain that plays.

Sustainable high performance, as described here, depends on teams doing two things at once: subtracting noise and restoring play. They create permission. the signal that it’s safe to be fully human “here. not just fully productive.” They protect space—actual cognitive room to exist outside performance mode. They also introduce a spark that starts with a belief that play can deliver unique value. keeping an opportunity to step outside comfort zones front of mind. then turning that belief into a prompt or call to action.

These three conditions are described as coming from interviewing and surveying nearly 1,000 people globally. The emphasis is on rituals and repeatable micromoments of human connection, exploration, and creativity rather than a six-figure transformation program.

A ritual doesn’t have to be complicated to carry real weight. One real-world example from one of Dara’s clients is “Kudos & Kinks.” In it. each team member gets 30 seconds to share one win and one thing that didn’t go to plan. It runs in every weekly team meeting. then receives a 15-minute slot at the monthly all-hands so it can be shared across teams.

The larger critique is blunt about how workplaces are built. The piece argues that schools reward compliance over curiosity, and that workplaces treat experimentation as risk. Over time, that has trained adults to see play as something children do—or a reward after real work is completed.

The bill for that decision is now arriving. denominated in engagement scores. AI adoption rates. and the roughly $10 trillion hole that the workplace data points to. The question, the argument insists, was never whether play works. It was whether organizations would be willing to take it seriously enough to make it part of every AI strategy.

No software purchase can close the gap created by a system that depleted and deprived the people expected to carry it forward.

AI adoption employee engagement Gallup workplace improvisation play leadership burnout neuroscience productivity