Dreams and reality
Is artificial intelligence the answer to western nations' declining immigration and ageing workforce, or just another challenge for policymakers? In this special guest article, our former Chairman Baroness Sarah Hogg explores this crucial debate, and asks if a happy balance is possible.
In the western world’s dull economic landscape of lilliputian numbers (low growth, low productivity, even inflation - hopefully - back down to low single figures), there is one statistic that has been changing dramatically: the level of immigration.
Especially in the UK. The coincidence of this big shift in the labour market with the impact of AI looking like perfect timing, but it requires policies tailored to the very different impact on different sectors of the economy.
The biggest “receiver” countries of the western developed world - the United States, Germany and the UK - all saw immigration rise in the past decade. But voters in all three have turned against open door policies, shifting under the weight of numbers away from the postwar consensus of obligation to those seeking asylum, as well as willingness to absorb large numbers of economic migrants. However, few of these voters fully appreciate the extent to which their governments have relied on the inflows to sustain economic growth and social services for an ageing population, or the extent to which other policies will now have to change. Is AI their salvation, or just another challenge to the policy-making capacity of western democracies?
The perennial puzzle: people vs policy
To start with the migration picture. For all the apparent similarity of the change in political mood in all three of these biggest people magnets, it is fuelled by some very different experiences.
The US, is, after all, an economy historically built on immigration. Until 1890, the inflow was led by Germans, followed by the English and Irish. Between 1890 and 1920, these were overtaken by the numbers from southern and eastern Europe, notably Italy, Austria, Russia and Poland. The third and greatest immigration phase, deemed to have started way back in the mid-1960s, but gathering pace in the twenty-first century, has come largely from Latin America and Asia. Mexico has been a prominent source, but immigration from Asia has most recently been strong.
The proportion of the US population that is foreign-born is still not, by international standards, particularly high. (It’s much lower than in, for example, Canada). Nor is it high by the US’s own historical standards, being much the same as it was in nineteenth-century America. But it is three times as high as it was in the early 1970s. And the recent numbers are big - 76 million immigrants in the past 60 years, 11 million in the past five years alone. Moreover, unlike in the UK, where despite the high visibility of channel boat crossings most of those arriving have been legal immigrants, in the US the number of “unauthorised” border-crossers has been high, and the first target of President Trump’s policies. Even if the brutality and lawlessness of his enforcement measures are rebounding against him, the political mood remains hostile to uncontrolled immigration.
Germany’s open-door approach has almost as deep roots, and the even today its new Chancellor insists that it will remain a “country of immigration”. In part that is a reaction against a rather different past. Germany was historically a country of emigration - especially to the United States - rather than immigration. But after the second world war, ethnic Germans fled eastern Europe for West Germany in their millions. A determination to welcome asylum seekers persisted throughout the 41-year life of that truncated German state, along with a somewhat less high-minded appetite for “gastarbeiter” - guest migrant labourers on formal working programmes - to fuel its economic miracle. This rapidly absorbed the less well-paid from other western European countries (Italy, Greece) as well as eastern Europe and Turkey.
Unification brought a further inflow from the east, together with a somewhat belated change in the treatment of gastarbeiter, allowing them to progress to permanent residence and citizenships. And an open approach to asylum seekers and economic migrants persisted through EU enlargement and the Syrian crisis in the 2010s, incidentally inflaming Brexit sentiment in a more fearful Britain. Russia’s invasion of Ukraine brought more refugees to Germany; the numbers of these still coming seems to have peaked, but with demand from elsewhere still strong Germany has not entirely escaped the change in political mood, and its general immigration rules have been tightened.
In the UK, the shift has been dramatic, both politically and economically. From a level of about 200,000 in the 2010s, dominated by entrants from the rest of the EU, after Brexit net immigration rose to a peak of 944,000 in the year to 2023, led by entrants for work or study from outside the EU. These legal inflows greatly exceeded the numbers struggling across the Channel, were a marked feature of Boris Johnson’s premiership, and helped the Brexit party to reinvent itself as Reform. But since 2023 the number has been dropping sharply: to a net inflow of 204,000 in the year to mid-2025, well below the ONS’s long-term forecast, and with some now predicting a net outflow in the next couple of years.
The impact on labour markets is already evident. But responses have been mixed. Germany has been perhaps the clearest in its policies, which are strongly focused on the continued attraction of the skilled and qualified. The US’s attitude to skills is less clear, with Trumpian hostility to highly-qualified tech-savvy Asian immigrants (particularly those in New York) clouding a picture supposedly focused on illegal entry of the unskilled.
Having tightened its rules, the UK keeps tinkering with them, with an approach not particularly well-calibrated to any part of the market. Business and industry complain of barriers to entry to scientists and engineers, in particular the mixed signals being sent to those capable of harnessing the capabilities of AI. Meanwhile employers of the lower-paid are suffering a complex mixture of labour market shortages and government-imposed increases in costs.
Some policy confusion may be understandable, because all these numbers should be treated with caution. Immigration figures are heavily subject to revision, while emigration statistics are little better than guesswork. But net immigration is now clearly falling, and at the same time as birth rates. The fertility rates in the major western economies are well below replacement levels; in the United States, for example, fertility has halved in just sixty years.
These rates can be expected to fall still further in the receiver countries, as immigrants, who typically account for a high share of births, also typically change behaviours in line with their host countries. In due course that leads inevitably to a shrinking labour force. In Europe, this demographic shift is already strongly marked. And as a result, Spain has been perhaps the most open in bucking the anti-immigration trend, to acknowledge that it will need newcomers to “fill the gaps”.
The ‘AI labourer’: can one size fit all?
Enter - superhero or demon king? - the AI game-changer. Economic liberator or job destroyer? It’s a debate in which the position of the UK becomes even more interesting. In a recent study, the International Monetary Fund singled out the UK as an economy particularly well-placed to benefit from the transformative power of AI[CH1] . This reflects the extent to which it has become a service economy, notably more so than Germany or the US, where indeed President Trump still seems obsessed with manufacturing.
In a policy-maker’s dream world, AI reduces the demand for human labour at just the moment when immigration is falling and babies are becoming scarcer. AI “fills the gaps” in the labour market. AI relieves white and blue collar workers of the dull parts of their jobs, which become more interesting and better paid. AI enables the UK to transcend the disadvantage of high labour costs, escaping the competitive pressure to lower these to Asian levels. Meanwhile government can take a free ride on these changes by taking credit for an increase in the minimum wage and increasing its tax take from employers.
And, of course, these dynamics are both possible and positive, and the advantages must be seized. But the real world is complicated by the fact that the impact of these changes is hugely different on different sectors.
The IMF study provides some simple clues. It estimates that 70% of the UK economy is open to the transformative power of AI, as compared with about 60% in the US, or in the developed world on average. Of course that’s a guess, but it’s an interesting one, and here’s where it gets tricky. This same study suggests that about half of this 70% is in “high-complementarity” areas, where people are needed as partners to AI or supervisors of its work. In other words, sectors where the transformative power of AI may make people operate more efficiently, but which will still need a whole lot of people. “Complementarity” requires more sophisticated policy-making than “transformation”.
In short, the service sector isn’t homogenous. For example, AI can transform legal processes such as conveyancing, doing them more cheaply (and quicker?) than expensive lawyers. The kind of generative AI that is now seen as pretty bog standard is probably good enough for this kind of thing, because it works on rules and past precedents, and that is how most law works too. But you wouldn’t, or shouldn’t, try to staff a care home with machines.
The National Health Service is vast enough to provide a case study the size of a whole small economy, with every extreme example of AI potential and human need. Robotic investigations and operations are transforming surgery, but an astonishing amount of information and communication is still paper-based, or siloed and non-transferable between institutions and even departments. Those examples give a glimpse of the potential. But on the other hand, demand for (human) nursing skills and beds continues to rise, and A&E patients pile up in corridors. AI can maybe help hospitals manage bed usage better, but it can’t staff hospital wards or prevent “bed blocking” by patients with nowhere else to go.
This illustrates the point that effective use of AI requires the development of new management skills; the effective use of labour resources is going to become more, rather than less, necessary as the use of AI increases. The political perception that the NHS is too full of managers may be blinding those responsible to its changing needs for skilled management, from the ward to the boardroom.
Changing patterns of migration have already been bringing some of these changes home to the NHS. Even before the tightening of entry regulations, some of the inflows on which it relied have been drying up. Demand for nurses from the Philippines, for example, has been increasing in Australasia, while the expansion of the Indian health service has made it harder to recruit there too. The main global migration outflows are now from Africa, or within Asia from Pakistan in particular, though the extent to which the US crackdown will divert Latin Americans to Europe is still unclear. All of which reinforces the message that UK policies need to be clear and well-targeted, if it is to match availability with need.
A digital partner for the people
The UK Government has been admirable in its focus on AI. But in self-congratulatory reports about the extent to which AI is being introduced across the economy, it makes the same mistake as those businesses which introduced staff pay incentives for AI use without actually measuring the impact. Now, however, most businesses are changing their approach, trying to develop complementarity, or “partnership”, between human and AI skills. Government, too, needs to take a sophisticated approach to AI policy-making.
In thinking about the economy in general, it isn’t good enough to suppose that resources will simply and efficiently shift to those sectors which are advantaged, because society still needs a lot of people-dependent services too. And not just in its public services. Society needs small people businesses beyond AI powerhouses in the south east. It needs cafes and hairdressers, artisan bakers, horticulture, craftsmen in garden sheds, all the little lifestyle businesses that fuel the lives of millions, as both producers and consumers. Some of these will be making brilliant use of AI, but all of them need a tax and regulatory framework that recognises their scale and essential humanity. Sticking-plaster policies, such as emergency tax subsidies for pubs, won’t do the trick; nor will a rash of council-owned community centres.
The struggle to revive high streets, on which successive governments have expended much effort, provides a key example of the need to avoid self-contradictory policies. There is no shortage of good intentions, and careful analysis of changing patterns of behaviour, while a good deal of government money has already been poured into community schemes. But the greatest threat comes from tax and regulatory changes that are closing small businesses in every high street in the UK - policies which can be managed by large, labour-shedding, AI-enhanced companies but not by those parts of the service sector that maintain a small, human, high street presence.
A failure to understand this can only result in serious sectoral and societal damage. Governments need to walk and chew gum at the same time - marching towards the embrace of AI while savouring small-scale, human entrepreneurship where it is most needed.
Guest Author: Baroness Sarah Hogg, Former Frontier Economics Chairman
