Everyone Is Worried About Entry-Level Jobs. The Real Casualty Is Middle Management.
The number that keeps circulating in workforce policy circles is 300 million — Goldman Sachs's 2023 estimate of jobs exposed to automation globally. Politicians quote it. Journalists illustrate it with pictures of call-centre workers and data-entry clerks. The entire public conversation about AI displacement has been built around a mental image of someone young, underpaid, and replaceable. That image is wrong. The workers being quietly gutted by this wave are not the ones at the bottom of the org chart. They are the ones in the middle of it — the team leads, the regional directors, the project managers, the people whose entire market value rested on being a conduit between information and decision-making. That layer is not being automated in the science-fiction sense. It is being made structurally redundant, quietly, one reorg at a time, and nobody is building policy to catch them.
What Middle Managers Actually Do — and Why It Translates Badly to This Moment
The conventional defence of middle management is that it involves human judgment, relationship-building, and contextual nuance — things machines supposedly cannot replicate. That was a reasonable argument in 2019. It is weaker now. A substantial portion of what a middle manager actually does on a Tuesday is synthesise reports, relay decisions downward, escalate problems upward, track project status, and sit in coordination meetings that exist because information does not flow cleanly between teams. McKinsey Global Institute's research on where workers spend time found that knowledge workers devote roughly 28 percent of their week to email and nearly 20 percent to searching for or gathering information. Those are not relationship-building activities. Those are latency costs — friction that accumulated because organisations lacked better tools. The tools now exist.
What is replacing that friction is not a robot with a clipboard. It is a flattened reporting structure powered by AI-assisted dashboards, real-time analytics, and large language models that can draft the weekly status update, flag the budget variance, and summarise the customer complaint thread before the VP even opens their laptop. When Shopify's CEO Tobi Lütke sent his widely-discussed internal memo in 2025 instructing teams to demonstrate why a task could not be done by AI before requesting headcount, he was not describing science fiction. He was describing a governance philosophy that makes an entire tier of coordination labour optional.
The Data Is There If You Look at Titles, Not Totals
Aggregate employment numbers obscure this because they are aggregates. The US economy keeps adding jobs — the Bureau of Labor Statistics headline figures look broadly stable — but the composition of corporate workforces has been shifting in ways that job counts do not capture. Layoff announcements from Meta, Google, Salesforce, and others in 2023 and 2024 disproportionately targeted what the companies themselves described as "management layers" and "redundant operational roles." Meta explicitly said in its Year of Efficiency communications that it was reducing the ratio of managers to individual contributors. That is not a coincidence. It is a strategy.
The Stanford AI Index 2024 noted that AI adoption is highest in precisely the industries — finance, professional services, information technology — that employ the largest concentrations of mid-level knowledge workers. When adoption is high and task automation is highest in those sectors, the exposure is not at the analyst tier, where entry-level workers are being supervised and taught. The exposure is at the tier doing the supervising — the layer that existed, in part, to manage information asymmetry that is now being resolved by software.
Why This Population Is So Poorly Served by Current Policy
Retraining programmes are almost universally designed for workers with a skills deficit. The community college coding bootcamp, the manufacturing-to-healthcare pathway, the digital literacy certificate — these make sense for someone who lost a job that required a narrow, automatable skill set and needs to acquire a broader one. They make very little sense for a 44-year-old operations director with an MBA, seventeen years of corporate experience, and a mortgage calibrated to a $140,000 salary. That person does not have a skills deficit. They have a structural demand problem. The job category they occupied is contracting, and no certificate course addresses that.
The strongest counterargument to this thesis is that middle managers will simply adapt — that the good ones will evolve into AI-augmented leaders who do more with smaller teams and remain valuable precisely because they can work at higher altitude. There is truth in that. Some will. The empirical question is what fraction, over what timeline, and against how many open roles. If the ratio of managers to individual contributors is being intentionally compressed, then even excellent managers face a musical-chairs problem regardless of their adaptability.
The entry-level worker who loses a data-entry job is visible, sympathetic, and politically legible. The 47-year-old regional sales manager whose position was "eliminated in restructuring" is invisible — too credentialed for safety-net programmes, too specialised for lateral pivots, too expensive to rehire at the same level. That invisibility is not a reason to feel sorry for them. It is a reason to look at them directly, because the displacement conversation that ignores them is not describing the actual shape of what is happening.
PROTECT YOUR INCOME NOW
If AI automation threatens your job, preparation is the only honest answer: know your risk, build a runway, and reskill before you have to. The Clankers 90-day programme structures exactly that — free, which is exactly why today is the right time to start.
Start the 90-Day Programme — Free →GET THE CLANKERS BRIEF — FREE
One email every Monday: the week's AI displacement moves that matter, and what to do about them. No spam. Unsubscribe anytime.