Why middle management isn’t ‘safe’ from AI displacement

Across companies in Singapore and beyond, strong performers are being let go as their roles get absorbed

Summarise
阅读简体中文版 (beta)
    • When a job is restructured out of existence, the worker loses the career infrastructure, professional identity and forward momentum that was built around the role.
    • When a job is restructured out of existence, the worker loses the career infrastructure, professional identity and forward momentum that was built around the role. PHOTO: BT FILE
    Published Thu, Jun 18, 2026 · 07:00 AM

    RECENTLY, a senior principal at a global advisory firm came to me after his role was eliminated. His company had just lost one big client, but his record was sound – he had also brought in two new projects that quarter.

    A few months earlier, another acquaintance who had spent a decade building a new division watched her company hire someone else to lead the function she created.

    Different industries, different high performers. Neither companies were struggling. In each case, the role itself had changed – completed, absorbed, or redesigned around the absence of the person who used to perform it.

    Artificial intelligence was never named as a reason. When a role is restructured out of existence, no one writes “replaced by an agent” on the redundancy letter. This might explain why the pattern has been hard to see.

    A “careerquake” is happening at scale, with structural shifts affecting individuals across firms.

    Meta’s high-profile retrenchment in late May made this visible, when more than 8,000 global employees were terminated amid announcements that the company had just committed to over US$100 billion in AI capital expenditure for 2026.

    When saving the worker isn’t enough

    The reality is that when a job is restructured out of existence, the worker also loses the career infrastructure, professional identity and forward momentum that was built around the role.

    The layer most exposed is not the C-suite. It is the cohort sitting just below strategy: programme directors, regional operations leads, transformation managers and senior analysts whose careers were built on translating decisions into action, coordinating across teams and holding operational coherence together across complex organisations.

    With the shift from AI tools to AI agents, the change is not incremental. While tools assist a person with a task, agents execute the workflow – coordinating, escalating, synthesising, reporting and following up without continuous human input.

    Those verbs map almost exactly onto the job description of middle management. Such roles demand judgment, institutional memory and the ability to hold context across competing priorities – capability requirements that increasingly do not disappear from the organisation when the person holding them does.

    It does not matter whether agentic AI can fully replicate what these people did. It is being deployed to absorb their roles regardless. The displacement is happening at the role level, rather than the task level.

    Singapore has recognised the direction of change. The Model AI Governance Framework for Agentic AI, announced at Davos in January 2026 by Minister for Digital Development and Information Josephine Teo, is the first of its kind globally.

    It is focused on system behaviour, including accountability, safety and transparency. What it does not yet address is what happens to the people whose jobs sit in the layer that agents are absorbing.

    In Singapore, the share of firms planning to hire in the next three months fell from 54.6% in February to 44.6% in March this year, a 10-percentage-point drop in a single month. PHOTO: BT FILE

    What the data misses

    According to the Ministry of Manpower (MOM), the labour market is holding.

    Overall retrenchments in the first quarter of 2026 held at 3,800 – rising slightly from 3,690 in Q4 2025 – with an incidence of 1.6 per 1,000 employees, within non-recessionary norms. Total employment grew for the 17th consecutive quarter.

    But look at the forward-looking data from MOM’s Labour Market Advance Release report, and a different picture emerges.

    The share of firms planning to hire in the next three months fell from 54.6 per cent in February to 44.6 per cent in March this year, a 10-percentage-point drop in a single month.

    Firms intending to raise wages fell from 39.3 per cent to 25.4 per cent in the same period.

    These indicators reveal what companies are already deciding, before it shows up in the retrenchment count.

    And the layer described in the opening examples is already moving. The retrenchment incidence rate of professionals, managers, executives and technicians hit 10.1 per 1,000 employees in 2025, well above the economy-wide rate of 6.3.

    According to MOM’s Labour Market Fourth Quarter 2025 report, retrenchments in this period were attributed not to poor business performance, but to reorganisation or restructuring.

    It also flagged “signs of restructuring that warrant continued monitoring”.

    The principal from the global advisory firm does not appear in that data as a warning sign. He appears as one restructuring decision in one quarter at one firm.

    So does the vice-president whose role was redesigned around her absence. Two separate decisions, each individually explainable, collectively invisible as a trend.

    The challenge is that role-level shifts are harder to see than job gains or losses. A role can be redesigned, absorbed or gradually narrowed long before it appears in retrenchment statistics.

    By the time the impact is visible in the data, the organisational change has often already happened.

    “Singapore has built a framework to govern its AI agents. We must also build visibility into what is happening to the people beside them.”

    Business as usual?

    Internationally, the scale of this shift is seen, but recognised as a move towards efficiency. A 2026 Gartner survey found that 80 per cent of companies already using AI are actively reducing headcount – doing so now, not planning to.

    A 2025 Korn Ferry survey of more than 15,000 professionals found that 41 per cent said their organisation had already cut management layers.

    The market consistently rewards these decisions as efficiency, not distress: When Coinbase announced a 14 per cent workforce reduction in May 2026, its share price rose 4 per cent.

    In Singapore, the signal is quieter only because the displacement runs role by role, quarter by quarter, inside reorganisations that each appear routine.

    The “careerquake” does not announce itself in the data. And so the systems designed to catch it – labour market monitoring, career support infrastructure, workforce policy – may yet be calibrated to address it.

    What needs to change

    Singapore has the institutional architecture to act on this before most countries do.

    MOM has already started tracking job redesign and AI-related role transformation.

    The next step is ensuring these indicators receive the same attention as headline employment and retrenchment figures, so policymakers, employers and workers can better understand how AI is reshaping work beneath otherwise stable labour market numbers.

    The country’s Economic Strategy Review has recommended earlier intervention in retrenchment and stronger transition support for workers affected by restructuring.

    It has also called for the proactive development of workforce transition plans before large-scale displacement happens.

    We also need companies to be transparent about AI deployment decisions. When organisations deploy agents to execute workflows previously managed by people, workforce impact assessments should come with an honest accounting: which roles are changing, who is affected, and what the company owes them on their way out.

    That is not a brake on AI adoption. It is the difference between restructuring that happens to people and restructuring that happens with them.

    Lastly, we need to rethink the support given to employees who need repositioning within a role category that is dissolving.

    Singapore’s career support ecosystem – SkillsFuture, National Trades Union Congress, Workforce Singapore – is built around skills retraining for known job categories and movement between existing roles.

    It also needs to cater for professionals whose work functions are being absorbed, and for whom no clearly defined destination role yet exists. The question they face is not “where do I find my next job?” but “what can I own now that an agent manages what I did?”

    Singapore has built a framework to govern its AI agents. We must also build visibility into what is happening to the people beside them.

    The time to act is now – not after the pattern becomes undeniable.

    The writer is the founder of Careerquake. She spent 25 years leading technology businesses across the Asia-Pacific and Japan. She now advises senior leaders navigating career and organisational transitions.