There was a time when organisations rarely used the word “human”. They spoke about labour, headcount, productivity and systems. People were assumed, present, necessary, but rarely named. Today, the word is everywhere: human-in-the-loop, human intelligence, human oversight, AI “beating humans”. When something must suddenly be specified so often, it is usually because it is no longer guaranteed.
That linguistic shift reveals a deeper anxiety. As artificial intelligence moves from experimentation into the core of business and government workflows, debate has focused obsessively on what machines can do. Much less attention has been paid to the more uncomfortable question hiding in plain sight: in an AI-driven economy, what exactly are humans for?
At the World Economic Forum’s annual meeting in Davos last month, International Monetary Fund Managing Director Kristalina Georgieva cautioned that AI could affect 40 per cent of jobs globally, rising to 60 per cent in advanced economies. Elon Musk, CEO of Tesla and SpaceX, speculated that AI could make “all jobs obsolete”, leaving humans to pursue meaning, or grow vegetables. Jamie Dimon, CEO of JP Morgan, proposed that governments need to intervene and prevent companies from dismissing people as technology advances, or at least slow the process down.
Three different tones, same underlying driver: the relationship between technology and work is approaching a breaking point. We have entered the man-machine age.
The WEF’s latest scenarios for jobs in 2030 give this unease a quantitative form. Fifty-four per cent of executives expect AI to displace jobs, while only 24 per cent expect it to create new ones. More than 40 per cent believe AI will raise corporate profits, yet just 12 per cent expect it to raise wages. In short, leaders largely agree that AI will make firms richer faster than it makes workers better off. That imbalance explains why the term “AI anxiety” is spreading beyond technology circles and into politics.
History offers some perspective as technological revolutions rarely transform work instantly. Electricity was commercialised in the 1880s, yet productivity gains took decades, not because the technology failed, but because organisations did not change. Factories, layouts and skills were designed for steam. Only when infrastructure, workflows and jobs were redesigned did electricity deliver its promise.
AI is following a similar pattern, though much faster. Adoption has surged – nearly 90 per cent of firms now use AI in at least one function – yet productivity gains remain modest. This is because organisations are automating tasks without rethinking work. AI is being bolted on to old processes rather than used to redesign how value is created.
This lag is not a flaw. It is a window. It gives societies time to shape outcomes rather than react to them. Every credible scenario shows that the future of work is not predetermined, but it depends on two variables that leaders can still influence: how fast AI advances and how prepared people are to work with it.
In the most optimistic scenario – a so-called “co-pilot economy” – AI progresses steadily, and workers are reskilled and broadly AI-ready. Routine tasks are automated, while humans move up the value chain. Human-machine teams become normal. Productivity rises gradually, job mobility increases, and demand grows for judgment, problem-solving, leadership and empathy. Work is reshaped, not eliminated. In this scenario, people have more fulfilling work and gain from the enormous benefits AI offers, from health to education, and from well-being to better work-life balance.
In the darker “age-of-displacement” scenario, AI advances faster than workforce readiness. Automation becomes cheaper than retraining. Entire occupational families shrink or disappear. By 2030, more than half of tasks in many sectors are absorbed by technology, rising far higher in exposed roles. Productivity rises, but unemployment, inequality and political strain follow. The wealth and income gaps increase rapidly between countries and within countries.
Crucially, in both scenarios, jobs are not equally vulnerable. Roles built around structured, predictable tasks – much back-office work, analysis, reasoning, administration and early-career white-collar jobs – are most exposed. Occupations rooted in human interaction, care, judgment and accountability remain harder to automate. A nurse’s work is far less affected than a junior analyst’s spreadsheet.
Labour-market data already reflects this shift. Demand for AI literacy is rising rapidly, while career ladders in many professions are being hollowed out. AI is not eliminating work evenly; it is reshaping who progresses and who stalls. This is where the debate is most evasive. The real challenge is not job protection, but human relevance. What we teach, how we teach it and how quickly adults can reskill will determine who survives – and thrives – in an AI economy.
Our education systems remain stubbornly front-loaded, designed for a world where learning ends early and work stays stable. That model is broken and must be redesigned, not simply improved. In an AI-rich world, learning must be continuous, modular and closely tied to changing skills: critical thinking, systems understanding, ethical judgment, creativity, collaboration and the ability to work productively with machines. Lifelong learning is no longer a social good; it is core economic infrastructure.
AI will not suddenly erase work. It will expose how poorly we have prepared people for change. For decades, we optimised organisations for efficiency and education for credentials, not adaptability.
The future of work is not primarily a technological problem. It is a human one. And whether the AI era becomes a story of shared prosperity or widening displacement will depend less on what machines can do than on how seriously we rethink education, skills and lifelong learning – before the window closes.
