Business & Deployment
Reskilling
Not prompting courses, but role redesign: clerical work becomes review work, creation becomes editing, execution becomes oversight. Reskilling means qualifying people for the activities that emerge alongside the machine — judgment, exception handling, system upkeep. Companies that take it seriously plan it like an operational changeover, not like a training series.
An example
A company sends its workforce to a half-day prompting course and ticks the topic off. Six months later hardly anyone uses any of it. What was missing was not knowledge about prompts but permission: no process where it was explicitly allowed to do things differently, no time to practice, no one to whom a failed result could be reported without looking foolish.
A closer look
What works is training that takes place on real cases and whose results flow back into the work. Courses without subsequent application fade within a few weeks. This is well documented and holds for AI as for everything else.
Why it matters
The scarce skill is rarely operation and almost always assessment — recognizing when a plausible-looking result is wrong. That cannot be taught in half a day; it develops through practice on real cases with feedback. An uncomfortable point follows: anyone who has never practiced the fundamentals themselves will find it hard to judge AI results. Entry-level work and apprenticeships are therefore where automation must be designed most carefully — not where it saves the most time.
Commonly misunderstood
The need rarely lies in operation. It lies in judging results, and that is learned only on cases where something is at stake.
Where this term does its work in the magazine.
1 article uses the term or takes up its principle.