Which jobs will be robotised before 2030 (and which won't, whatever you've heard)
Every few weeks a headline claims robots will wipe out X million jobs. It almost always comes from an exposure study, and it almost never explains what that kind of study measures. It is worth stopping there, because the gap between what gets published and what the data say is large, and it affects real decisions by real people.
What an exposure study actually measures
The International Labour Organization puts it with unusual clarity for an international body: exposure indicators should be interpreted as early signals of possible change, not as predictions of job displacement. What these models do is take job descriptions — static, written years ago — and calculate which tasks a machine could perform. They do not calculate whether it pays off, whether the firm will buy it, whether the collective agreement allows it, or whether the robot exists. They measure what technology could do, not what employers will implement.
The surprise: not the jobs you'd expect
Popular intuition says routine, low-skill work gets automated first. The more recent AI research points the other way: higher-skilled cognitive roles — business, finance, computing, maths, education — show up among the most exposed. The reason is mechanical. A language model is cheap to deploy over office work, while a robot that manipulates objects in a messy environment remains expensive and fragile. Physical and cognitive automation are not advancing at the same pace, and the physical side is far behind.
The figure that does hold up: 58% of employers
The World Economic Forum's Future of Jobs 2025 found that 58% of employers expect robotics and autonomous systems to transform their operations by 2030. Note the verb: transform operations, not cut headcount. That is the difference between a warehouse reorganising how boxes move and a warehouse laying off half its staff. Historically, in logistics and manufacturing, the former has been far more common than the latter — though the content of specific roles does change.
What is really pushing this: demographics
One force rarely makes the headlines and explains industrial interest in robots better than any other: Europe is ageing. The population aged 65 and over goes from 21% to 29% before 2050. In warehousing, construction, industrial cleaning and care work there is no surplus of workers looking for a job — there are jobs nobody wants to fill. In those sectors the robot does not arrive to replace someone, it arrives to cover a shift that has been vacant for months. That is the argument actually moving money.
The roles that will change before 2030
With the data in hand, the short list is fairly consistent: material movement in warehouses, palletising and end-of-line, repetitive welding, machine tending, inventory checking, large-surface cleaning, inspection in hazardous environments and internal transport in hospitals and factories. All share three traits: a controlled environment, a repeatable task, and a cheap consequence when the robot gets it wrong. Remove any one of the three and automation stalls.
The ones that won't, whatever you've heard
Clinical diagnosis, emotional care, exception handling and final authority over a decision. This is not sentimentality, it is the structure of the work: these are tasks where the unusual happens constantly and where being wrong is expensive. A robot that handles 95% of cases and fails on the remaining 5% does not save a job, it creates a new one that consists of supervising the robot. In many facilities that is exactly what has been observed: more reliability, maintenance and engineering staff, not fewer people overall.
How to read the next headline
Three questions filter out most of the noise. Does the study measure exposure or actual job losses? Is there a robot deployed today doing that task, or is it a trade-show prototype? Is the robot's total cost per hour, including maintenance and downtime, below the human shift? If the headline answers none of the three, it is probably not describing the 2030 labour market — it is describing somebody's funding round.
