In Part 1, I split AI into two kinds: Displacing AI, which replaces a person in a task, and Elevating AI, which makes a person more capable. The obvious next question is how to tell them apart.
Most people assume it comes down to the technology. In my experience, it rarely does.
Deployment decides
The same tool can land in either category depending on how you deploy it. Take one large language model. Set it up one way and it replaces your copy editor. Set it up another way and it makes that same copy editor twice as fast. Same model, opposite outcomes. The deployment decided.
That is where leadership comes in. Nobody chooses “AI” in the abstract. You choose a configuration, a workflow and a rollout, and those choices decide whether a tool pushes people out or pulls them up. So this is a decision you own, not a technical question to hand to the engineers.
The four question test
When a client is weighing an AI investment, I ask four questions.
One: what happens to the human role? After launch, does the task run without the person who used to do it, or does that person stay and reach further?
Two: who stays accountable for the output? When responsibility stays with a person, you are usually looking at elevation. When it disappears entirely, you are looking at displacement.
Three: where does the gain show up? Is the saving measured as fewer people for the same output, or as more output per person? The two look similar on a slide and are very different in practice.
Four: who feels the effect? This is the one people miss. A single rollout can lift a senior worker while displacing a junior one, so the answer can differ by role within the same project.
These questions won’t hand you a verdict, but they give you a direction, and that is usually enough to change the conversation. You stop debating “AI” as one big thing and start deciding what you want it to do to the people it touches.
Displacement is a choice
They also remove a common excuse. Leaders sometimes talk as if displacement simply happens to them. If a deployment removes people, someone chose that, for reasons that can be examined. Seeing it as a choice is the first step to making a better one.
So before you argue about whether AI is good or bad for your organisation, run the four questions and find out which kind you are actually building. Then match your workforce plan, governance and messaging to the answer.
In Part 3, I will take on the biggest myth about where all this is heading: the idea that population size decides a country’s AI future.