Three in four business decision-makers say AI has made individuals faster at their jobs, according to data from Miro. The businesses they work for aren’t necessarily moving any quicker.
The lone genius problem
The gap, says Miro’s Chief Evangelist Pete Bradd, comes down to how AI tools get adopted inside most teams: one person at a time, rather than as something shared.
“The idea that breakthrough work comes from one brilliant person working alone is an old myth that AI is bringing back to life,” Bradd says. AI tools are typically built for individual use, he says, so people end up working faster and more independently than before, but every prompt, agent and custom workflow ends up belonging to just one person. “The better the tech gets, the more one person can achieve without anyone else,” he says. “But because this adoption is fragmented and out-of-sync rather than aligned, the gains stay siloed.”
It’s not a talent problem, Bradd says. It’s what happens when people don’t share the prompts, context or decisions behind their work. That knowledge stays with the individual instead of becoming something the team can build on. “This results in duplicated effort, inconsistent approaches, and more time spent reconciling work between team members than AI ever saved,” he says.
Bradd points to MIT research on collective intelligence to back up the pattern: a group’s performance tracks more closely with how well its members coordinate, their social sensitivity, and how evenly they contribute, rather than with the average or highest individual IQ in the room. “The same logic applies to AI,” he says. “Having more lone geniuses doesn’t make a smarter company, it creates a company full of smart people who can’t build on one another’s work.”
Spotting it in your own team
The clearest sign that AI gains are staying with individuals rather than lifting the whole team, Bradd says, is a mismatch between what gets reported and what’s actually happening. “If self-reported time savings are increasing while project timelines, handoffs and delivery speed at the team level are unchanged or even slowing, it suggests that individual efficiency isn’t necessarily translating into collective performance,” he says.
A few concrete warning signs to watch for: duplicated work, where someone rebuilds a report from scratch because they didn’t know a colleague had already done it; handoffs that take longer than they used to, because the person receiving the work has to reverse-engineer what was done and why before they can use it; and a widening gap in how effectively different people on the same team use AI, which leaves everyone unsure what’s already been done or whether it can be trusted and reused.
“Ultimately, the question is whether AI is reducing the total effort required to move work from idea to outcome, rather than simply reducing the time one person spends on a particular task,” Bradd says. “If individual task completion is speeding up but collaboration, quality control, decision-making and handoffs are becoming more complicated, that’s a strong sign that productivity gains are remaining at the individual level rather than improving how the team works as a whole.”
Closing the gap
Bradd’s advice for teams starts small and specific rather than sweeping. “Pick two or three recurring processes and agree as a team on how AI fits into each one: what it’s used for, what a good prompt looks like, what still needs a human decision,” he says. Standardising a handful of the most important workflows, rather than trying to standardise everything, gives the team a shared baseline without slowing people down or killing individual initiative, he says.
From there, the work needs to move into shared tools rather than stay in someone’s personal AI chat history. “If the work already lives in a shared canvas, doc, or workflow tool, that’s where the AI-assisted steps should happen too, not off to the side in someone’s personal ChatGPT history where nobody else benefits,” he says. Keeping the AI-assisted steps inside the team’s existing shared workspace means the context and reasoning behind the work stays attached to it, rather than disappearing when that person moves on to something else.
The last step, Bradd says, is the simplest and most overlooked: talking about how the work got done, not just what came out of it. “A quick ‘here’s the prompt/workflow that got me here’ turns one person’s shortcut into the team’s shortcut,” he says. “This is what actually closes the gap between individual speed and team performance: the capability moves from person to person until it’s just how the team works, not what one person happens to be good at.”