The AI backlash, Work and the Risk of Social Breakdown

By Graham Vanbergen: I have written many articles about the breakdown of the globalised world we live in. In the European Financial Review, in 2016, I warned about a rising corporatocracy, the following year about the crisis of trust in democratic institutions, in 2018 about the duel between big-tech and government, and surveillance capitalism a year later. This creeping degradation of society almost appears to be a trajectory – one that is soon to explode in our faces.

The public argument about artificial intelligence is still being framed by the people who stand to profit from it. Big Tech tells us AI will raise productivity, cure administrative waste, create new industries and free human beings from drudgery. Some of that may be true. But it is not the whole truth. The missing half of the story is darker: millions of people may discover that the economy no longer needs them before politics has worked out what to do with them.

This is not science fiction. Goldman Sachs has estimated that the equivalent of 300 million full-time jobs globally are exposed to automation. The IMF says around 40 per cent of global employment is exposed to AI, rising to about 60 per cent in advanced economies. The World Economic Forum expects huge churn by 2030: tens of millions of jobs displaced, even if new roles are also created. The reassuring phrase is always “transition”. But transition for whom? A 52-year-old administrator, call-centre worker, legal assistant, bookkeeper, designer, junior programmer or translator may not experience transition. They may experience redundancy. The the generation entering the workplace, this future reality is already looking grim. They are already getting angry.

The great danger is not simply that AI destroys jobs. It is that it destroys the social contract around work. For generations, ordinary people accepted capitalism’s harsh edges because the bargain was clear enough: work hard, earn wages, pay bills, raise families, and perhaps move forward. AI threatens to sever that link. And it’s not so bad if you are in your late 50s or 60s, but for almost everyone else who is younger with aspirations, it is a direct threat. If companies can expand output while shrinking payrolls, then productivity gains will flow upward to owners of platforms, data, chips, cloud infrastructure and intellectual property. The worker becomes a cost to be eliminated. The consumer, however, remains necessary. That is the contradiction Big Tech does not want to discuss: who buys the product if millions lose income?

This is where AI becomes a macroeconomic risk. A company replacing 10,000 workers may please shareholders. A thousand companies doing the same may weaken demand across the whole economy. Households without secure wages spend less. Small businesses lose customers. Local high streets decline. Tax receipts fall. Welfare bills rise. Governments borrow more. Then the very businesses that celebrated automation face a poorer, angrier customer base.

The social consequences could be severe. Communities already damaged by deindustrialisation know what technological abandonment feels like. Closed factories produced more than unemployment; they produced despair, family breakdown, addiction, resentment and political extremism. AI could do something similar to white-collar work. This time the shock may hit office towns, graduates, creatives, middle managers and service workers who once thought automation was something that happened to other people. And don’t forget why we have Brexit and the complete fracturing of the political two-party norm in the UK and Donald Trump in the USA.

Data centres in Ireland account for approximately 22 per cent of the nation’s total metered electricity usage. The backlash has already begun around them. Communities are protesting against the power-hungry, water-hungry physical infrastructure of AI. Residents object to rising electricity demand, water use, noise, land grabs, pollution and tax breaks for companies already worth trillions. These protests matter because they reveal something important: people increasingly see AI not as a magical tool but as an extractive system. It takes land, water, energy, data, public subsidies and human work – then returns private profit.

 

Revolutions begin when people believe the system is rigged, contemptuous and deaf

 

That anger could widen. If AI is seen to enrich a small technological elite while hollowing out employment, politics and democracy itself will not remain stable. The 2008 financial crisis, austerity and globalisation already taught voters that elites can make mistakes while ordinary people pay the price. AI may intensify that lesson. A worker who loses a job to software may not be comforted by lectures about innovation. A town losing both employment and affordable electricity to data centres may not be impressed by speeches about the future.

Could this lead to violence or even revolution? It would be irresponsible to predict revolution as inevitable. But it would be equally irresponsible to dismiss the risk. History shows that mass economic insecurity, visible inequality and political humiliation can produce unrest. Revolutions rarely begin because people are poor alone. They begin when people believe the system is rigged, contemptuous and deaf. AI could create exactly that perception if the benefits are privatised and the costs socialised.

The danger is especially acute because AI threatens status as well as income. Work is not only a wage. It is identity, usefulness, routine, dignity and belonging. A society that tells millions they are “obsolete” should not be surprised if they react with rage. People may protest, sabotage infrastructure, support extreme parties, attack corporate offices, resist automation in workplaces, or demand punitive taxation of AI firms. Political fractures could open between the employed and displaced, the young and old, graduates and non-graduates, cities and regions, humans and machines.

The inequality problem is central. AI rewards scale. The best models require enormous computing power, vast datasets, specialised chips and access to cloud infrastructure. That favours the largest technology companies and the richest economies. Smaller firms may become dependent on rented intelligence from a handful of suppliers. Workers may become dependent on systems they do not own and cannot challenge. Nations without AI infrastructure may become customers rather than competitors. The result could be a new form of technological feudalism: a small class owns the machines; everyone else rents access or competes against them.

There is also a democratic risk. If AI reduces the bargaining power of labour, unions weaken. If unions weaken, wages stagnate. If wages stagnate while asset owners become richer, politics polarises. People begin to seek scapegoats. Migrants, welfare claimants, civil servants, foreign competitors and “the establishment” all become targets. AI may not create authoritarian politics by itself, but it can provide the economic fuel: insecurity, anger and distrust.

Governments are badly behind the curve. Politicians talk about skills, but reskilling is often a slogan masquerading as policy. Not everyone can become an AI engineer. Not every displaced worker can move city, retrain for years, or accept lower-paid care work. And if AI systems also improve in coding, design, legal drafting, analysis, customer service and management, the ladder of “new jobs” may keep moving away just as people try to climb it.

The welfare question is unavoidable. If AI makes large numbers of people redundant, who pays them? Universal basic income is often suggested, but it would require vast taxation and political consent. Taxing AI profits, compute power, data extraction or automated labour may become necessary. But Big Tech will resist fiercely, warning that regulation will kill innovation. Governments desperate for investment may compete to offer subsidies, cheap energy and tax breaks, even while their own citizens face job losses.

This is the political trap: states may fund the infrastructure that undermines their tax base. Public money, public grids, public water and public tolerance are being used to build systems that could reduce public employment and private wages. If that bargain becomes visible, the backlash will be explosive.

None of this means AI should be banned. The technology has genuine value in medicine, science, education, accessibility, engineering and public administration. But societies must stop pretending that efficiency is automatically humane. A civilisation is not successful because it can produce more with fewer people. It is successful if people can live with dignity, security and purpose.

 

The question is who gains, who loses, who decides and who pays

 

So what can be done. The first requirement is honesty. Companies deploying AI should disclose likely labour impacts. Governments should monitor job displacement in real time. Workers should have consultation rights before AI systems are introduced. Public procurement should favour human-supporting AI, not human-replacing AI. Tax systems should capture windfall automation gains. Competition authorities should prevent a handful of firms from owning the intelligence layer of the economy. Energy and water approvals for data centres should require community consent and transparent public benefit.

Above all, AI must be treated as a political economy problem, not merely a technology story. The question is not whether machines can do more. The question is who gains, who loses, who decides and who pays.

If AI becomes a tool that helps nurses, teachers, engineers, small businesses and public services, it may be remembered as a great advance. If it becomes a machine for transferring wages into monopoly profits, it may become the defining social conflict of the century. Big Tech is selling the dream of abundance. But unless governments act, many people may experience AI as dispossession – and dispossessed people do not remain quiet forever.

 

 

 

---