AI and the Future of Invisible Work: Automating Hidden Jobs

When people picture AI taking jobs, they tend to picture robots on factory floors or algorithms writing news articles. These images are not wrong — but they miss the more pervasive story. The disruption already underway is quieter, less visible, and in many ways more consequential: the systematic automation of the labour that keeps organisations running but never appears in any job title, strategy deck, or annual report. It is the work between the work — and it is disappearing faster than almost anyone has noticed.

The Jobs Already Being Transformed

Consider the actual composition of a knowledge worker's day. Research by McKinsey and RPA vendors consistently finds that between 30 and 60 percent of working time in office roles is consumed not by the nominal function of the role — advising clients, developing products, managing people — but by the operational layer that surrounds it: scheduling, data gathering, report formatting, email triage, meeting transcription, document review, compliance checking. None of these activities appear in any prestigious job description. All of them are being automated, right now, at scale.

Enterprise AI tools have reached sufficient maturity that this automation is no longer a pilot programme or an aspiration — it is in deployment. Microsoft Copilot summarises email threads and drafts responses. Notion AI generates meeting notes from transcripts. Salesforce Einstein qualifies leads and suggests next actions. Harvey drafts legal documents and research memos. Each tool addresses a specific slice of the invisible operational load, and each reduces the number of hours a human needs to spend doing work that feels like administration rather than value creation.

Automating the Invisible: What Is Actually at Stake

The sociologist Craig Lambert coined the term "shadow work" for the unpaid, unacknowledged labour that has quietly migrated to consumers — checking yourself in at the airport, building your own IKEA furniture, managing your own investment portfolio online. AI is now performing a similar migration within organisations, absorbing the shadow work of professional roles. When that work disappears from human schedules, several things happen simultaneously.

Productivity per person increases. Headcount requirements decrease for the same output volume. The nature of remaining roles shifts toward higher-complexity, higher-judgment tasks. For some people, this is genuinely liberating — suddenly freed from hours of operational drudgery to focus on work that demands and develops real expertise. For others, the operational tasks were precisely the accessible entry points into a profession — the junior work through which skills and institutional knowledge were acquired before being trusted with greater responsibility.

Opportunities and Disappearance: The Honest Reckoning

The honest answer about digital transformation and employment is that both net creation and net destruction are happening — in different industries, at different skill levels, and on different timescales. The World Economic Forum's 2025 Future of Jobs report estimates 85 million jobs displaced by automation over five years, alongside 97 million new roles created. The numbers look reassuring until you notice that the displaced roles and the created roles are not in the same places, do not require the same skills, and are not accessible to the same people.

How to Prepare: For Individuals and Organisations

The most durable career positioning in an AI-automated environment is to become expert at the things AI does poorly: contextual judgment in genuinely novel situations; building trust with specific humans in specific relationships; creative synthesis across domains that have not been connected before; and the metacognitive skill of knowing when to trust an AI output and when to interrogate it.

For organisations, the imperative is redesigning roles and career ladders around the new reality rather than simply removing the automated tasks and calling what remains a full role. The junior positions that previously provided learning-through-doing need to be reconceived — perhaps as AI oversight roles that still provide exposure to the full range of professional work, even if the hands-on execution is increasingly assisted. Companies that neglect this will find themselves in three years with a workforce that lacks the institutional experience that only comes from doing things, not just reviewing them.

The future of work in an AI-shaped economy will not be determined by the technology alone. It will be determined by the choices organisations, policymakers, and individuals make about how to deploy it — and whether those choices are made with clarity about who bears the costs of the transition and who captures the gains.