I am getting tired of the phrase “AI will change work.”
Not because it is wrong.
Because it is too easy.
Of course AI will change work. The more interesting question is how.
The papers I have been reading all point in the same direction: the future of work is becoming a role-design problem.
Not just “which jobs disappear?”
That question is dramatic, but incomplete.
The better question is: how do roles change when AI enters the workflow?
Middle managers become translators between strategy, employees, customers, and AI systems. Leaders become AI-supported sense-makers, but still have to do the human part. Workers need to know when to trust the machine and when to push back. Domain experts become interpreters of AI outputs. Customers may interact with hybrid agents where the boundary between human and machine is not always obvious.
Everything becomes a little more blended.
And blended work needs design.
Otherwise, people end up with strange half-roles. The human is responsible but not empowered. The AI recommends but does not explain enough. The manager is accountable for a system they do not fully understand. The customer is told help is available, but nobody is sure whether help means bot, human, or emotional obstacle course.
This is where organisations need to grow up quickly.
AI is not just a tool added to existing jobs. It changes the shape of the job.
What should humans still own?
What should AI handle?
What needs escalation?
What needs judgment?
What needs care?
My reflection is this: the future of work is not humans versus AI. It is role design under pressure. Companies that get this right will not simply automate tasks. They will redesign responsibility, learning, trust, and decision-making around the new reality. Much harder than buying software. Much more important too.