In late 2024, the Pew Research Center surveyed more than 5,000 employed Americans and found that 52 per cent were worried about how AI might be used in their workplace. Roughly a third expected fewer job opportunities for themselves over the long run. Worry, in that data, clearly outweighed excitement.
Most coverage takes a number like that and runs it straight into the obvious story: people are afraid the machines will take their jobs. The fear is real, and the numbers behind it are not small. But it is worth pulling apart two worries that usually travel together, because the louder one tends to hide the deeper one.
The first is about income. The second is about everything else a job quietly carries — the standing it confers, the skill it took years to build, the sense of being good at something, the feeling of being needed. One is a worry about a salary. The other is a worry about a self.
The fear is real, and widespread
None of that worry is irrational. The economic risk is genuine, the pace is unusual, and a worker does not need a grand theory to notice that a tool doing part of their job cheaply is not obviously good news for them. Any honest discussion has to begin by granting that the fear is reasonable on its own terms.
But the unease is not only about pay packets.
Why a job is never only a job
One of the clearest accounts of this comes from the social psychologist Marie Jahoda, whose mid-twentieth-century work on unemployment has aged remarkably well. Jahoda argued that paid work supplies far more than money: it gives people a structure to the day, contact with others beyond the family, a sense of shared purpose, regular activity, and — most relevant here — status and identity. Her central point was that losing work damages wellbeing beyond what the drop in income alone can explain. People out of work, in her account, tend to lose their sense of time, their self-respect, and their place in things. The pay packet was only one of the supports a job had been quietly holding up. If that is right, then a technology that threatens the work threatens those other functions with it. Think of the senior analyst who has spent twenty years learning to read a balance sheet the way a sommelier reads a wine list, now watching a chatbot produce a passable version of that judgement in nine seconds. The fear is not only that the paycheque goes. It is that the scaffolding goes with it.
Competence is a need, not a vanity
A second piece of well-established psychology fits the same shape. In self-determination theory, developed over decades by Edward Deci and Richard Ryan, competence is one of three basic psychological needs, alongside autonomy and a sense of connection to others. People are not only chasing outcomes; they need to feel effective at things that matter to them.
When that need is frustrated, the theory holds, the result is a felt sense of ineffectiveness that wears on wellbeing. So the prospect of a system that performs, in seconds, a task you spent a decade learning to do well is not a trivial bruise to the ego. It presses on something the research treats as a real human need.
This is what the word “useless” is usually pointing at. When people say they are frightened of being made useless, they are not only talking about employability. They are describing the loss of a feeling that carries its own weight.
What the “just reskill” answer misses
This is where the workplace conversation tends to go wrong. The standard reassurance — offered by companies and commentators alike — is that AI will not replace people but augment them, and that anyone worried simply needs to reskill.
As career advice, that is often sound. As an answer to the fear, it frequently misses the mark. If part of what someone values is the competence and standing they have already built, then being told to set that aside and become a beginner again is not a comfort. It is a fair description of the loss they were afraid of in the first place.
It also explains a pattern that is easy to misread inside organisations. When experienced staff resist a new AI tool, the temptation is to file it under fear of change, or plain stubbornness. Sometimes that is all it is. More often it is something more specific and more defensible: a person protecting an identity and a hard-won expertise that the tool, however useful, quietly discounts. Worth noticing before reaching for the word Luddite.
The harder thing to say
So here is the harder thing to say. The fear is legitimate, and the “just reskill” answer is glib — but neither of those facts entitles anyone to a permanent right to the work they used to do. A self built almost entirely on a single professional competence was always a fragile self, and the arrival of capable machines is exposing that fragility rather than creating it.
Organisations that pretend otherwise, soothing their senior staff with vague reassurances, do them no favours. The kinder move is to name what is actually being lost, help people grieve the specific competence rather than the abstract role, and then push on. Treating fear of AI as a failure of nerve misreads it. Treating it as a veto misreads it just as badly.
What it deserves is honesty, not accommodation.
