New research on workplace AI adoption has found something that should give small business owners pause before they roll out the latest chatbot or copilot tool company-wide: the employees getting the most value from AI already had an advantage before the software showed up.
The advantage in question isn't access to better tools or more training time. It's job design. Employees in roles with more autonomy, more complex decision-making, and more control over how they structure their day are extracting far more value from AI tools than employees in rigid, tightly scripted roles. In plain terms, AI is amplifying the gap between good jobs and average jobs rather than closing it.
This matters because it cuts against a narrative that took hold a couple of years ago. Earlier studies, including widely cited research on AI-assisted customer support agents, found that lower-performing or less-experienced workers often saw the biggest jumps in output when given AI assistance — essentially using the tool as scaffolding to perform more like their top-performing peers. That research fueled optimism that AI could be a leveling force inside companies, helping newer or less-skilled employees close the gap with veterans.
The new findings complicate that picture. Rather than a uniform leveling effect, the benefit appears to depend heavily on the structure of the job itself. Workers with more discretion over their tasks can experiment with AI, figure out where it helps, and fold it into judgment calls. Workers in highly standardized, closely monitored roles have less room to do that, even if the tool is technically available to them.
Why it matters
This fits a broader pattern seen with earlier waves of workplace technology, from spreadsheets to enterprise software: new tools tend to reward people who already have flexibility and decision-making authority, and offer less to people whose jobs are defined by rigid procedure. AI adoption inside companies has largely followed the same script — tools get deployed broadly, but the people who benefit most are self-selected based on how much latitude they have to use them creatively.
What this means for small businesses
For a small business, this is a planning problem, not just a tooling problem. Handing every employee a ChatGPT or Copilot license and assuming productivity gains will spread evenly across the staff is unlikely to match what's actually happening on the ground. The employees with more autonomy — managers, senior staff, anyone whose job involves judgment calls — will likely pull ahead faster than hourly or process-driven staff.
That creates a practical choice. Owners can either accept an uneven adoption curve, or they can deliberately redesign some rigid roles to give frontline employees more decision-making room alongside the AI tool, which research suggests is what actually drives the benefit. The second option costs more management time upfront but may be the only way to get value out of AI tools for the majority of a workforce rather than a handful of already-strong performers.
There's also a quieter risk worth naming: if AI gains concentrate among employees who are already doing well, pay and performance gaps inside a small team could widen even as overall productivity numbers look good on paper.
What to watch
Watch for how companies measure AI's return on investment over the next year. If firms start reporting productivity gains at the team or department level rather than company-wide, that's a signal the uneven-adoption pattern is showing up in their own numbers too. Also watch whether vendors selling workplace AI tools start marketing role redesign or workflow consulting alongside their software, which would suggest they've noticed the same gap.
The bottom line
Rolling out an AI tool to the whole staff does not guarantee the whole staff benefits equally; job structure, not just access, appears to determine who actually gains ground.