Every year, the Office for Budget Responsibility tells us the same story: based on reasonable assumptions, our nation’s long-term public finances are unsustainable. As an ageing population leans heavily on the state pension, the NHS and social care, keeping tax and spending on autopilot would lead to government debt-to-GDP doubling again to nearly 200 per cent of gross domestic product within four decades.

There are no easy policy fixes that avoid some “austerity”. A surge in working-age immigrants might buy us some time, at best. A burst of unexpected inflation can chip away at our debt mountain but cannot “inflate away” future promises of hip replacements and a triple-locked state pension. Throw in a few recessions and future pandemics and we’re staring down the barrel of higher taxes and spending pressures for decades to come.

Or are we? For years, I’ve said that only a significant boost to economic growth could realistically save today’s pay-as-you-go welfare states. Enter artificial intelligence. A recent survey from The Kent A Clark Center for Global Markets at Chicago Booth, the business school, found that 46 per cent of top American economists agreed that AI could supercharge economic growth rates in the United States and western Europe over the next 20 years (only 4 per cent disagreed). Surely, that’s a godsend to HM Treasury?

It’s easy to see why these economists are excited. AI theoretically has the potential to kick productivity into overdrive. Just ask Shakked Noy and Whitney Zhang, economists at the Massachusetts Institute of Technology who discovered that average productivity for writing tasks got a massive boost when writers used ChatGPT. AI-driven automation could streamline business processes, unleash a tsunami of new creative content and eventually build a “virtual workforce” of assistants to deliver more projects, economic activity and so tax revenues.

Healthcare and education, two stagnant public services, could be revolutionised, putting downward pressure on government spending, too. AI tools might offer bespoke lessons for individual children. Existing AI technologies already could slash healthcare costs by 5 per cent to 10 per cent, according to American economists, with savings found through streamlined administration, harnessing clinical data and medical procedure efficiency improvements. That’s aside from the non-fiscal benefits, such as improved healthcare quality and reduced stress on clinicians.

It’s very possible, then, that AI will radically ease the fiscal constraints that we’ve felt already for 12 years. Yet should policymakers assume this will happen? Absolutely not.

First, as Nick Bloom, the British economist, reminds us, real GDP growth has been slowing since the 1950s, despite huge technological innovations. Any GDP uplift from AI-driven growth will still meet the headwinds of demographic trends, stalling global trade, Britain’s land-use planning noose and vested interests in our nationalised public services that will seek to prevent or delay AI adoption.

Next, the internet has taught us that groundbreaking technology doesn’t guarantee measured GDP booms. Computers and the internet transformed our lives, yes, but a lot of that involved improvements in the value of our leisure time. During much of the Industrial Revolution, growth did not take off stratospherically even though the foundations for prosperity were being laid. No wonder 50 per cent of economists are “uncertain” that AI will boost growth substantially and few would endorse bolder predictions of growth rates doubling or more.

Finally, AI itself might change the supply and demand for government spending. Politicians might use efficiency savings to expand their footprint into new areas or to fund pricey, cutting-edge AI treatments in the NHS. AI itself will create private sector winners and losers, and some job losses, which many will demand governments react to by spending more on support and retraining.

Jeremy Hunt said recently: “I truly believe that AI can be the answer to our productivity and growth challenge.” It could well be, making future chancellors’ jobs easier. It is prudent, however, to budget on the basis that none of that is guaranteed.