Avoiding a two-class workforce
In every company that adopts AI, a gap opens quickly. A few people use the tools intensively, produce much more, and get noticed. Others use them little or not at all, fall behind, and start to be seen as a cost. Left alone, that gap hardens into two classes of employees, with very little movement between them.
Why it happens by itself
The people who adopt early tend to be the ones who were already confident, already well placed, already given the interesting work. AI multiplies their advantage. The others often lack not ability but time, permission or someone to show them. Without a deliberate effort, the company rewards the first group, promotes from it, and slowly writes off the second.
The result looks like efficiency but behaves like a caste system: an inner circle that decides and an outer ring that executes and worries. Morale drops, knowledge stops flowing, and the people in the outer ring leave or stop trying.
Building the bridge
Rotate people through the new way of working. Everyone spends time in the redesigned processes, not only the volunteers.
Pair the advanced users with the others on real work, not in classrooms. Learning happens on actual cases.
Keep promotion open. A meaningful share of the people who move into the new, higher-value roles should come from outside the early-adopter group. If they all come from the same circle, the bridge is not working.
Keep the entry path alive, so new hires can still learn and grow.
Fund it from the savings. The training and pairing are part of the redesign, paid for by what it saves, not an extra to cut when budgets are tight.
What to measure
Three simple indicators, reviewed by management every month: the gap in usage between teams, the share of promotions coming from outside the early-adopter group, and who is leaving voluntarily. If the people leaving are concentrated in one group, you already have two classes.