Barclays announced this week that it will cut approximately 5,000 jobs as the bank increases its operational reliance on artificial intelligence. The redundancies are concentrated in back-office functions — compliance processing, trade settlement, routine customer service, and documentation review — where large language models and automated decision systems are now performing work that previously required substantial human teams. The bank described the restructuring as part of a strategic shift toward AI-augmented operations that would, it said, improve efficiency and reduce error rates in high-volume processing tasks. Barclays is not alone: across global banking, Goldman Sachs, JPMorgan, and HSBC have all either announced or implemented similar programmes, and industry analysts estimate that the financial services sector could automate between 20% and 40% of current roles within a decade.
The received wisdom
The centre-left and progressive response to AI-driven job losses follows a now-familiar script: the disruption is real, the pain is unevenly distributed, and the answer lies in robust retraining programmes, universal basic income pilots, and stricter regulation of AI deployment to ensure human oversight. Unions representing financial sector workers have argued that Barclays and its peers are using AI as cover for cost-cutting that serves shareholders at workers’ expense. There is also a fairness dimension: the jobs most at risk in banking are concentrated among lower-paid, more diverse sections of the workforce — operational and administrative staff rather than traders and executives. The implicit bargain of the knowledge economy, that education and adaptability offer reliable employment security, is under stress.
A different read
The progressive diagnosis of AI disruption contains genuine insight, but its preferred remedies have a poor historical track record and involve a degree of magical thinking about the state’s capacity to manage technological transitions. Let us take the honest version of this argument seriously before examining its limits.
The automation of back-office banking is not abstractly different from the automation of textile production in the eighteenth century or of assembly-line manufacturing in the twentieth. Each wave of labour displacement caused real hardship — hardship that was not adequately anticipated, cushioned, or addressed by the institutions of the day. The left’s historical critique of market capitalism’s treatment of displaced workers is well-founded. The industrial revolutions were brutal in their transitional phases, and the political consequences — Luddism, Chartism, the Labour movement itself — were direct responses to the failure of markets and governments to manage displacement humanely.
Where the progressive prescription falters is in its assumption that government-directed retraining can reliably bridge workers from declining to growing sectors at the pace that technology moves. The evidence from post-industrial Britain and the American Rust Belt is dispiriting: retraining programmes have a poor completion rate, tend to retrain workers for roles that are themselves being automated, and cannot overcome the geographic mismatch between where good new jobs are created and where displaced workers live. The call-centre worker in Bristol whose job has been replaced by a chatbot does not become a prompt engineer in six months of night school.
More fundamentally, resisting the automation itself is not a viable option, and governments that have tried — whether through digital services taxes designed to disadvantage AI investment or through procurement rules requiring human staffing minimums — have typically generated perverse incentives without preserving jobs. The competitive pressure is global: if Barclays does not automate, it loses ground to Frankfurt, Singapore, and New York banks that do. This is not a defence of the outcome; it is a recognition of the constraint.
What a right-leaning analysis adds to this conversation is a scepticism toward two orthodoxies simultaneously: the market-fundamentalist view that disruption is self-correcting and the progressive view that the state can manage technological transitions through sufficiently ambitious programming. The honest position is that AI-driven displacement in finance will create genuine losers who will not be made whole by either market forces or government intervention on the timescales that matter to their lives. The appropriate response is not to slow the technology — which would fail — but to be clear-eyed about the distributional consequences and to design fiscal policy accordingly, including through the tax base that AI productivity gains generate.
Britain’s forthcoming Budget, already under pressure from elevated inflation and borrowing costs, will need to address this without the luxury of fiscal expansion. That is the genuine policy bind — not whether AI is good or bad, but how to share its productivity gains in a period of constrained public finances.
What to watch
- Whether the UK government’s forthcoming AI regulation framework addresses workforce displacement directly, or focuses solely on safety and governance concerns
- The pace at which Barclays’ redundancy programme triggers similar announcements from HSBC, Lloyds, and NatWest, which face the same competitive pressures
- Whether any political party uses the Barclays news to advance a concrete tax-on-automation proposal, and whether that proposal is costed
- The next Bank of England quarterly assessment of financial sector employment trends, expected in Q1 2027
--- J