AI is Destroying the Job Market

The UK employment market is sending out a message that neither job applicants nor, for different reasons, politicians want to hear. New figures from Adzuna released earlier this week point to an extraordinary collapse in graduate vacancies, with only around half as many jobs available today as there were a year ago; just 8,383 graduate roles were listed on the site in July.

This poses an obvious question: where exactly are new graduates supposed to go to find work? The political response thus far has largely consisted of hand-wringing about skills, about whether the education system is sufficiently subservient to the demands of employers, and supposed overdiagnosis of mental illness. But it is one thing chivvying young people onto the ladder; quite another when that ladder is suddenly yanked up out of their reach.

Economic weakness and uncertainty undoubtedly play a role, but the widespread adoption of AI is clearly the big, hulking elephant in the room. Inevitably, when the chance arises, employers look to boost profit margins and cut labour costs where they can, whether through automation, outsourcing or offshoring. The advent of AI must be like all their Christmases come at once. They find themselves presented with a once-in-a-lifetime opportunity to slash their largest single cost: wages.

For decades, employers increasingly demanded that applicants be served up to them ‘job ready’, and so the costs of training them increasingly passed from themselves and on to the state, and to students themselves in the form of loans. Hence the huge expansion of higher education in the 2000s and 2010s; young people were assured that the way to progress in life was to take on ever-increasing levels of student debt (which they could worry about paying back later) on the understanding that this would provide them with a stable white-collar career and a solid foothold in the middle class.

Now we see that bargain breaking down. Employers know that large language models can handle many of the tasks previously allotted to entry-level employees at a mere fraction of the cost; after all, LLMs don’t require salaries, pensions, sick pay or promotions, nor do they take holidays, or go on maternity or paternity leave.

Graduates are by no means the only ones affected by this. People with decades of experience now report submitting dozens or even hundreds of job applications before securing an interview, let alone an offer. Many roles once considered relatively secure are now easily automated, either in part or out of existence altogether: marketing, copywriting, admin, accounting, basic coding, and numerous others.

Optimists, of course, insist that just as earlier technologies replaced obsolete jobs with new ones, so will AI. But this is not a law of nature and such assurances ring hollow when AI companies openly brag about how many people they’re going to throw on the scrap heap.

The Luddites get a bad press from bourgeois historians, but far from being purblind primitivists, they were correct that the factory system posed a mortal threat to their livelihoods and that any new jobs it provided would likely not be on offer to them. At least their children and grandchildren had cotton mills to work in. What happens now, with no obvious source of alternative employment waiting to absorb displaced workers?

Politicians, flattered by a phalanx of tech industry lobbyists, refuse to face up to this because it would force them to address questions around who controls the technology and whose interests it serves. After so many years of economic stagnation, they are also desperate to have some growth to brag about and hope that AI can deliver it, but they remain evasive, at best, about how the proceeds of its much-vaunted productivity gains (primarily a euphemism for sacking people) will be distributed.

So what we get instead is another media moral panic about alleged shirkers on benefits, with right-wing columnists fulminating about the rising social security bill and centrists wringing their hands, but reluctantly reaching similar conclusions. Some tech oligarchs talk airily about bringing in some form of universal basic income as a salve, but decades of scapegoating have created a political climate in which entrenched anti-welfare attitudes mean the prospect of others receiving ‘something for nothing’ is anathema to many.

In any case, a UBI introduced as an afterthought, or sticking plaster, for mass AI-induced unemployment would probably end up as a miserly stipend for those workers who find themselves discarded, their skills no longer valued and their bargaining power destroyed, or as just one more subsidy enabling businesses to get away with paying low wages. Needless to say, a few work experience placements at Sainsbury’s won’t cut it as a response either.

AI is exacerbating existing inequalities of wealth and power, but the genie is already out of the bottle. Rearguard actions by organised labour to resist or regulate its adoption might slow its advance – which is not insignificant – but are unlikely to reverse it. The political debate, however, is miles behind what is already playing out in the real world. It is surreal and insulting to see politicians, and equally cosseted pundits, pontificating about the importance of acquiring skills and experience just as their market value is being gutted.

For socialists, the core demand must ultimately be that AI, as with other technologies, is deployed in the service of the many, not the few, and that its proceeds and capabilities are socialised, rather than being captured by the oligarchs who currently control it. AI, built as it is on industrial-scale theft by big tech, is the product of humanity’s collective intellectual endeavours, accumulated over countless generations; any benefits it offers therefore ought to be the rightful, common property of humanity as a whole.