Alvarez & Marsal's talent practice lead warns AI pilots often test the wrong thing, and premature cuts create talent debt

An inconclusive pilot stops being neutral once a workforce plan is waiting on it. Campbell's explanation for why so many finish that way is narrow and specific. The pilot was set up to test whether the technology works, when the question that matters is whether it solved a business problem ^[2].

A job is a bundle of tasks, but people also fill roles inside an organisation's social system, and Campbell's argument is that the social system is the part of work AI cannot see ^[3]. Matt Alder, who hosts the show, opened the episode on the same point. "When companies start talking about AI, conversations often focus on headcount, as if a job was just a list of tasks," he said ^[4].

The five A's model Alvarez & Marsal uses stages AI's impact as avoid, assist, augment, automate and autonomous ^[5]. Sorted by what happens to the person doing the work, two of those five stages describe AI running the work without a human in the loop, two keep a person in it, and one removes the work from the organisation before automation is in question ^[6]. A plan that books every AI-relevant task as a saving has collapsed all five into one.

There is a second measurement gap in the same argument. Employees who bring their own AI tools capture the individual benefit while the organisation carries the risk and gets no commercial return ^[7]. Campbell's advice is to engage and scale that use instead of policing it ^[8].

The talent debt claim is the load-bearing one, and it is the thinnest part of the record. Campbell says cutting too far, too fast in pursuit of AI savings creates a talent debt, and that some big tech companies which cut early are now slowly rebuilding the capability they lost ^[9]. The episode carries that claim and leaves the specifics out: which companies, how big the rebuild, over what period ^[15]. A managing director who advises on workforce strategy and organisation design has an interest in telling clients to redesign before they cut ^[1]. The claim is still checkable in a way the episode does not check it: named firms, rehiring for capability they removed, on a stated timeline.

The timing asymmetry makes this a sequencing problem. A cost target is delivered this quarter; capability that has to be rebuilt surfaces over years, and the executive who signs the reduction is often not the one who funds the reversal. Campbell expects the question to move from what organisations should do with AI to whether they are designing work around it, and he says the ones that pull ahead will redesign the work, the roles and the social fabric together ^[11][12].