The Economist has spotted a threat from artificial intelligence that isn't mass unemployment.... AI is very, very good at helping people complain. Tax appeals, benefit reconsiderations, planning objections, parking tickets, grievance letters — anything that used to require a wet Sunday afternoon, a ring binder and a functioning grasp of administrative law can now be produced to professional standard in about 30 seconds.
For the individual this is unambiguously good. For the machinery of government it is, apparently, a catastrophe.
Inertia was never a bug
The revealing part of the argument is the admission buried inside it: our bureaucracies were built for the age of the post and the telephone, and have always depended on a certain amount of systemic friction to function at all.
Strip away the euphemism and what it means is that the state's budgets, staffing levels and service standards have been quietly calibrated on the assumption that most people entitled to something will not claim it, and most people wrongly refused will give up. Take-up rates for pension credit have hovered around two-thirds for years. Vast numbers of disability benefit refusals are overturned on appeal — by the minority who appeal. That was never a design triumph. It was a subsidy extracted from the confused, the exhausted and the unwell.
AI hasn't broken the system. It has removed the tax on articulacy that was holding the system together.
The numbers are already in
The employment tribunals are the canary. Single claim receipts rose 39% in the year to March while disposals actually fell 12%, leaving 64,000 open cases — a 55% jump. Some five-day hearings in London are now being listed into 2029. Judges have gone on the record linking both the volume and the sprawling complexity of submissions to claimants drafting with AI tools, and note pointedly that there is no sign of the economic downturn that would ordinarily explain a surge in disputes.
Applications for interim relief — an obscure, rarely-used remedy — have risen more than a hundredfold. That is not a wave of new injustice. That is a wave of people discovering procedural options that were always theoretically available to them and practically available only to those with a solicitor.
Pruning rights is not a plan
Which is why the proposed remedy deserves more scepticism than it has received. Close the loopholes, prune procedural rights, simplify the appeals systems. Note the direction of travel: the citizen has become better at using the rules, so we shall have fewer rules.
Some of this is genuinely sensible — Byzantine appeal architecture serves nobody, and an AI-armed state answering AI-armed citizens is at least symmetrical. But "prune procedural rights" is a phrase that should never be allowed to pass without someone asking whose. Rights that only functioned because most people couldn't exercise them were, in a meaningful sense, not rights at all. They were decorative. Removing them now, precisely at the moment they threaten to become real, is not administrative housekeeping. It is a rationing decision dressed as a modernisation programme.
Final thoughts
There is a version of this story where a state discovers it has been under-delivering for decades and responds by fixing the delivery. There is another where it responds by making the entitlements smaller. Nothing about the last fifteen years suggests we should be confident which one we're going to get.
