Companies can fall into an easy trap: if AI is strategically important, everyone must use it. But the importance of a technology does not make every tool universal.
Different roles do different work. Where AI removes routine or speeds up analysis, the benefit can be significant. Where those tasks barely exist, it becomes another browser tab.
Start with tasks, not people
Instead of asking “who is still not using AI?”, ask which operations repeat, consume time, involve text processing, information search, classification, or preparing alternatives.
The task, not the job title, should be the first unit of analysis.
Not every time saving is worth the rollout
If a tool saves someone five minutes once a week, training, licensing, data controls, and support may cost more than the benefit.
That is not an argument against AI. It is normal automation economics.
Mandatory usage distorts behavior
As soon as prompt counts or active-user ratios become KPIs, people learn to create activity.
You get AI meeting summaries nobody reads, unnecessary text, pointless generations, and reports proving how successful the rollout is.
Different roles carry different risk
For one team the main question is speed. For another it is confidentiality, accuracy, legal responsibility, or the cost of an error.
That is why a single rule that “everyone must use AI” is almost always too crude.
Scale proven value
A better path is less exciting: find a concrete task, measure the effect, define data rules, understand limitations, and only then expand the practice.
In one function AI may become a daily tool. In another it may remain an occasional assistant. That is fine.
Not using AI can also be rational
If an employee does not use AI, it does not automatically mean resistance to change. The tool may simply not improve their current work enough to matter.
The goal is not to turn everyone into an AI user. The goal is to improve work where AI can actually improve it.