The dominant narrative around AI and employment goes something like this: artificial intelligence will progressively learn to do your job, and one day, it will replace you. This is a compelling story, neatly packaged for headlines. It is also, for the most part, the wrong story. The real mechanisms through which AI is destroying jobs are far less photogenic - rooted not in the technology itself, but in the perception of what it might become, and in the financial engineering that perception enables.
Consider the unfolding saga of Paramount's bid for Warner Bros. Discovery — a deal now valued at around $82 billion. As Scott Galloway observed on a recent episode of Prof G Markets, when you overpay for an asset at this scale, the arithmetic demands that you find ‘efficiencies’ on the other side. The deal will be heavily leveraged. Debt must be serviced. And the primary lever available to service that debt is labour cost - which is to say, people. AI provides the narrative cover. The acquirer doesn't announce mass redundancies, it announces a "transformation programme" underpinned by artificial intelligence. The language of innovation softens what is, in essence, a very old playbook.
This is not unique to media. Grant Thornton's global private equity tracker reveals the increasingly dominant role of leveraged buyouts across sectors that were previously insulated from this kind of financial intervention - housing, retail, healthcare, professional services. The logic is always the same: acquire, leverage, extract efficiency, return capital to a few limited partners. AI has become the newest and most compelling justification for the "efficiency" part of that equation, which has always been code for reducing headcount.
But here is the critical nuance. It is extraordinarily difficult for AI, in its current form, to replace an entire job. A job, however, is a collection of tasks. AI can replace individual tasks with increasing competence. The question is not whether AI can be an architect, a lawyer, or a financial analyst, but whether AI can perform enough of the tasks within those roles to fundamentally alter the economics of employing a human to do them. We are not there yet, but the perception that we are, or soon will be, is doing real work in the world.
Sinead Bovell articulated this precisely on a recent episode of her podcast. Some companies, she noted, are not laying off employees because their jobs have been automated. They are laying off employees to become leaner in anticipation of a transformation they believe is coming. Others are cutting staff directly to redirect capital toward AI infrastructure. In both cases, the driver is perception — a bet on a future that has not yet arrived.
This perceptual pressure operates on employees as well as employers. If you believe, rightly or wrongly, that your employer could plausibly replace elements of your role (or your whole job) with AI, your bargaining power diminishes. You become more compliant, less likely to negotiate, less likely to push back. The mere possibility of replacement disciplines the workforce without a single algorithm being deployed. This is not a side effect. For many organisations, it is the point.
And as robotics, world models, and physical automation mature, this phenomenon will migrate beyond white-collar work. The same financial logic - acquire, leverage, automate, extract - will apply to warehousing, logistics, manufacturing, and service work.
AI, then, poses an enormous threat to job security globally. But the threat operates primarily through perception and financial engineering, not through the one-for-one replacement of workers that dominates the public conversation.
The real engine of disruption is not the technology. It is the story we tell about the technology, and the capital structures that story enables.
