Summary: this article diagnoses the UK government’s current science and technology policies, including both strengths (commitment and ambition in many fields of science) as well as weaknesses, partly the result of the outsize influence of key elites. It then shows what a more convincing agenda would look like complementing existing moves on topics like AI with policies on ten key fields, including technology adoption, regional inequality, public engagement, and the intersection of technology, trade and security. It argues for taking seriously basic political questions about values and about who benefits, which are sidelined by the crude ‘more-ism’ that currently dominates policy.
……………………………………………………………………………………………………………………………………………………
One of the many mysteries of this Labour government is its science and technology policy. Peter Kyle is amongst the brightest and most impressive Labour ministers, fluent in the media, widely respected and apparently on top of his brief. His department, DSIT, is ambitious and active, particularly around AI, and closely aligned with the broader focus on economic growth.
But despite plenty of activity, and many good initiatives, Kyle has yet to give his department a convincing political direction. Many argue that his department’s policy stance has been captured by small, London-based science elites, which means that it has little to say on many of the big issues of science and technology policy, and even less that brings a centre-left or social democratic perspective, as opposed to continuity with the Conservatives.
DSIT was set up by the Conservatives and its approach under both parties can best be summarised as a return to ‘more-ism’: the view that more science and more technology are goods in themselves.
This view was common in the mid-20th century. But it became ever more problematic over the last half-century thanks to everything from DDT and Thalidomide to nuclear proliferation and cloning, Internet impacts on children to climate change, biased AI, and a hundred other issues. The smarter scientists have long recognised that the idea of simply ‘following the science’ is rarely a good guide to action. More-ism came to be seen as a naïve anachronism that ignored the complex and uneven impacts of technology on gender, class, age and place and the most basic political question - who benefits?
But this questioning of ‘more-ism’ was never accepted by much of the London-based science elite who have been remarkably successful at blocking debate. That elite includes the Royal Society, the Wellcome Trust, convenors like the Foundation for Science and Technology, and more recently the Tony Blair Institute, and figures such as former science minister David Willetts and the current science minister Patrick Vallance, and the leaders of most recent reviews, notably Paul Nurse.
More recently, the biomedical elite has been joined by the influence of Silicon Valley which now – remarkably – counts three former party leaders as advocates (Nick Clegg, William Hague and Tony Blair) and has been adept at lobbying for minimal regulation and taxation. Key figures have taken powerful roles in government, including in charge of UK competition policy (perhaps symptomatic of a government that has recruited many business lobbyists – rather than entrepreneurs or innovators – into powerful positions). The big consultancies, all closely tied to Silicon Valley, exercise as big a behind-the-scenes influence as ever. Peter Mandelson’s role as a lobbyist for Palantir who now, as UK Ambassador to the US, acts as a broker/promoter for Palantir’s growing role in the NHS and defense, is a particularly striking example.
These elites contribute often high-quality analysis. They include many impressive and serious individuals. They are good advocates for curiosity-driven research and have prompted many useful initiatives, including on Metascience, and they’ve done a good job in pushing the government to take AI very seriously (rightly in my view).
But they are the representation of interest groups, not always of the public or the national interest. They have narrowed the debate within government rather than widening it and often blurred the divide between objective advice and self-interest. And amidst the futurism, they have an oddly old-fashioned view of science policy that doesn’t always fit well with many of the priorities of the late 2020s and the UK’s interests at a time when the world order is changing very fast.
Their perspectives are far preferable to the anti-science alternatives of Trump and others (in the US the anti-science trend amongst Republican voters has been underway for thirty years; so far this hasn’t infected the UK but that could easily change, which is one reason why the issues I cover below are so important).
It’s not hard to understand why a new government wanted to benefit from some of the authority of the science elite, especially one with so little scientific expertise in its higher reaches: Darren Jones and Lucy Powell seem to be the only members of the Cabinet with any science or technology background, in both cases as undergraduates.
But dependence on small elites risks blinding the government to many of the most important challenges facing the UK let alone the challenges of steering the next generations of AI, quantum, synthetic biology or geo-engineering – and, as I show later, attachment to ‘more-ism’ puts it at odds with mainstream public opinion.
In what follows I describe some of the challenges facing DSIT and science policy more generally, why they are blind spots now, and what a more rounded approach would look like. Less than a year after the election, there is still plenty of time for government to rethink and appreciate that it has options.
1.Technology Adoption as well as R&D is crucial for growth
The first, and most visible, blind spot involves how to mobilise science and technology for growth, the most obvious challenge for the government given stagnant productivity and GDP growth.
There’s a healthy, and desirable, emphasis on keeping the UK on the cutting edge of technologies including AI and life sciences. Unfortunately, this leaves policy very incomplete. Because the science elite is so concerned with the frontiers, it has never shown any interest in questions of technology adoption. Yet how well manufacturers, shops, schools, or repair shops adopt new techniques, from data-driven marketing to AI, almost certainly matters as much to UK productivity as high- end R&D.
For decades, however, this has been seen as either irrelevant or low status by the science elite. It was ignored by literally all the major reviews of recent years and hardly mentioned in most Labour policy documents which explains why there is still a glaring absence of policies and institutions to address it (a review has at least been started though nothing has yet resulted).
That should be a big concern for the government. It is a glaring hole in strategies for growth, and this makes it even more likely that the next few decades will repeat the pattern of recent history, when technological change helped to widen inequalities.
But it should also worry government that the science elite has largely ignored decades of research challenging the traditional ‘linear’ assumption that the supply of new scientific knowledge from basic research is the main source of innovations - a belief that is obviously untrue in most UK industries, from finance and media to services, but also only partly true in fields like the life sciences.
Decades of study have shown the limitations of this linear model, and the equally flawed assumption that commercialisation and spin-offs are the key means through which R&D drives growth. Instead experience points to the far bigger importance of flows of people (particularly PhDs) and of innovation ecosystems with the depth of knowledge and relationships that’s key to technological prowess.
This choice to ignore these lessons is reflected in the balance of spending between science and technology; and between the southeast and the rest of the country, and it’s reinforced long-standing failures to attend adequately to the ‘D’ of R&D, to engineering and to technology.
This is not to say that the UK should ignore the frontiers of science and technology. That would be perverse. Rather a coherent strategy needs to give equal weight to the frontiers and to the challenges of adoption.
2.Tackling the decline in science productivity
If more resources are being committed to science and technology, as they have been by recent governments, then it’s vital these deliver value for money, especially when money is tight. Yet one of the most glaring trends of the last 50 years has been declining returns to R&D, starting with the pharmaceutical industry (it's sometimes called ‘EROOMs law’, the mirror of Moore’s law which describes the steadily rising productivity of digital technologies).
Because scientific productivity has been in long-term decline, some have argued for tackling it head-on, suggesting that it may have roots in the conservatism of arrangements for funding and organising research. The remedies may include innovations in all of these, as well as more use of AI in research.
But this issue too is never mentioned in the many reviews and policy statements of the last decade. Advocates of ‘more-ism’ struggle to accept that ever more money has been delivering less in terms of results, and this is an uncomfortable topic for incumbents with backgrounds in the biomedical field where the problem first became apparent.
John van Reenen, who chairs Rachel Reeves council of economic advisers, is fully aware of the issue – as one of the authors of a seminal article documenting declining productivity in R&D. The paper - ‘Are ideas getting harder to find’ from the American Economic Review - presented ‘evidence from various industries, products, and firms showing that research effort is rising substantially while research productivity is declining sharply’. This is encouraging some to apply the scientific method to science itself, using RCTs and other experimental methods to test out what actually works in creating and sharing new knowledge, and accelerating the use of AI in research. It’s one reason why ARIA is very much to be welcomed - bringing new methods, people and mindsets and encouraging bold, radical ideas.
But the incumbents of the science elite have resisted applying the scientific method to their own methods and policies. I once asked a senior official in the Department of Health if any scientific or objective method was used to allocate public health research funds – they told me it’s almost entirely about individual relationships and who has access.
In short, on the issue of declining productivity too, there are no signs that either the diagnosis or the prescriptions have been much debated let alone accepted.
And when it comes to institutions the world of science is as unthinkingly conservative as tech industries are bold and creative. Recent reforms drew on very old models creating ARIA (based on DARPA, which was founded in 1958); Catapults (based on Fraunhofer Institutes which were founded in 1949); and UKRI, consolidating a funding council model that dates back to 1913 (in Medical Research). There may be good reasons for this conservatism – but those involved confirm that alternatives weren’t even considered.
3. Connecting science to place – not just in the south-east
In a country still struggling with big regional inequalities, there are bound to be pressures to ensure that more people, and more regions, can play a full part in science and technology.
This is rarely straightforward. The presence of a university doesn’t automatically have any impact on growth or opportunities in a region. But no local growth strategies can afford to ignore knowledge-intensive industries, and the benefits of attracting high tech investment, labs or research centres.
So far, the government has had little to say about this. The one exception is the move to deepen an innovation arc between Oxford and Cambridge, a potentially very dynamic growth zone.
This is overdue and to be welcomed (as a resident of Luton I am biased in support). But it’s odd that this now appears to be the only big place-based idea in play. London, Cambridge, and Oxford have traditionally had a disproportionate share of public spending (and of influence in Whitehall), even though private R&D investment is more evenly spread across the country.
Sustained public investment in the innovation economy has delivered little for much of the UK population: trickle-down hasn’t really happened, which should worry policy-makers and politicians.
One part of the shift that’s needed is a broader integration of spatial policy and science, for example through investing more heavily in the UK’s many innovation districts as growth hubs. This doesn’t mean spreading money everywhere. But it does mean focusing on a small number of growth poles and not just one.
4. Engaging the public – not seeing them as bystanders
One recent poll found that barely one in ten of the British public believe that publicly funded R&D benefits them, and there are plenty of signs that not just that misinformation, conspiracy theories, and anti-science could in time become a serious obstacle to science-led growth, but also that the public might question the privileged position of science in public spending decisions. Various past governments saw the value of opening up the conversation about science, from cloning and GM crops to AI.
But this seems to be off the agenda in the UK now and we have reverted to the view that the public should be grateful observers, not participants. For many years science policy makers assumed that the priority was to educate the public – more understanding of science would make them more supportive. This turned out to be far too simplistic and many countries have shifted towards a more interactive, conversational approach. But the UK seems to have regressed.
As an example, although the public are fairly warm about the potential of AI, a large majority don’t trust AI companies: yet current policy essentially ignores them in favour of lobbying from the industry.
It's striking that the pressures from the creative industries on intellectual property in AI seemed to come as a surprise to the government, as did the pressure from parents on regulation of social media platforms. These concerns simply don’t figure in the ‘more-ist’ perspective.
I currently teach a course on the practicalities of AI in public services and one of the key issues we look at is how to avoid scandals over the next few years thanks to badly managed AI (as has already happened in the Netherlands and the US). Careful public engagement is at least part of the answer (rather than capture by the consultancies who currently seem almost as powerful as the science elite), as is attention to detail.
It’s worrying how much discussion of AI over-generalises with over-heated hype often from leaders with only a dim awareness of how AI actually works. AI means very different things in medical diagnosis, passports, tax collection, policy design, forecasting, transport planning and customer services, to mention just a few examples.
Having tried (and failed) to cancel Horizon in the late 1990s when I worked in No 10, I find it all too easy to imagine equivalents in the late 2020s. The government repeats flaky claims from consultancies about likely savings from AI in public services, as if nothing had been learned from the wholly wrong predictions similar consultancies made in the past. I’m strongly in favour of much more use of data and AI in government. But naïve boosterism makes mistakes more likely.
5. Smarter regulation rather than just less regulation
Next comes the question of how to reshape regulation to fit the priorities of the government and of the public. The more-ists often default to a ‘less-ist’ position on regulation, seeing regulation primarily as a brake on innovation. Recent months have seen a lot of rhetoric about slashing regulation, which some have criticised as a kind of ‘karaoke Thatcherism’ (though to her credit Thatcher realised you need regulation to stop monopoly and promote competition and innovation and she created quite a few regulators).
There are good reasons for wanting to both cull and modernise regulation, and I support the government’s move to set up a Regulatory Innovation Office (having advocated exactly this kind of ‘anticipatory regulation’ over the last decade). But there are two obvious problems with the current position.
One is that although the public in principle will support streamlining regulation, they will quickly oppose any moves that go too far and subordinate the public interest to private interests. Just closing or merging regulators without a clear view of how to weigh up the benefits and costs of rules is bound to lead to problems. The UK used to be more sophisticated on these issues – for example, when the Human Fertilisation and Embryology Authority was set up in the 1990s with a dual purpose of ensuring public legitimacy and scientific exploration.
A second problem is that the decision to put the RIO in DSIT may turn out to have been a mistake, since it has little power over regulation (and unsurprisingly others are now vying to control this space). It’s not hard to imagine how this could unravel and there’s a risk that the recent appointment of David Willetts to run the Regulatory Innovation Office, an able former Conservative science minister appointed by a former Conservative chief science adviser, now science minister, will turn out to be more a symptom of the pattern I’ve described than a solution to it.
6.Science, society and not wasting talent
For the longer term one of the crucial questions is how to make the most of the UK’s creative potential. Good STEM courses in schools and degrees at universities are part of the answer, but not enough, especially for a government concerned about social mobility.
A recent study analysed over a million creators of patents in the US and found that children born into the richest one per cent of society were ten times more likely to be inventors than those born into the bottom 50 per cent. The most likely innovators had parents who worked in technology or were brought up in areas with strong technology industries, and so had early experience of working with gadgets, fiddling with TV sets, radios or computers. Smart kids from poor backgrounds weren’t getting the right support or experience – their talent was being wasted (the title of the research was ‘Lost Einsteins’). The researchers showed that US innovation could quadruple if women, minorities, and children from low-income families became inventors at the same rate as men from high-income families. But this, they argued, required shifting investment from subsidies for venture capital (mainly benefiting relatively affluent men) towards funding programmes that would expose girls and boys to innovation, and the practice of science, at an early age.
Very similar patterns seem to hold in the UK – potentially worsened by the narrowing of the curriculum over the last decade (despite many good programmes to encourage children to learn coding). Again, however, we have heard nothing from DSIT about how to widen participation in the high-tech economy, in part because this has never been a topic of much interest to the elite.
This particular gap also reflects the apparent absence of any thinking about social innovation. In much of the world, including the EU, Japan, Canada, Brazil and countries like Malaysia or Singapore, it’s increasingly taken for granted that social innovation is a vital complement to technological innovation. As an example, it’s widely accepted that many of the barriers to net zero are now social ones rather than technical ones, and most of the UK’s other missions depend every bit as much on social innovation as new technology. Yet in this respect, too, the UK has gone backwards, with apparently barely a murmur of debate in the government.
7.Connecting STI to trade and the new geopolitical realities
In the new age of tariffs and geopolitical competition an obvious challenge for the UK now is how to align trade, technology, and regulatory policy. But it’s still not clear where the capacity to do this well sits, and what methods should be used. I have been in many Whitehall sessions grappling with this, and the analytic methods which could be used. But it remains hard to see either what is being done or who is doing it.
The issues are very complex and becoming more so as battlelines harden over digital, data, AI, and EVs. The science elite worldview doesn’t fit well with the shape of the world of the late 2020s, with weaponised national corporatism increasingly the dominant model in the US and China.
The Silicon Valley evangelisers have also become ever more problematic as the US, Europe and China pull apart. Is the UK content just to be a client for Microsoft, Palantir and others? Should we develop our own competing ‘stack’, or should we increasingly try to align with the EU as it edges towards greater digital sovereignty?
The need for a much tougher-minded mobilisation of regulation and trade to serve national interests is already becoming obvious. But it’s telling that the recognition of this in the recent government report on AI, which committed to creating a new ‘Sovereign AI’ capacity, seems to have come from the security services rather than from politicians, scientists, or political advisers. And elites with backgrounds in the biomedical field on the one hand, and close to a no-longer-reliable ally on the other, may not be well-placed to lead a shift to more emphasis on security and weapons, and a sharper focus on national interest.
Another difficult security question is also fast becoming unavoidable. If defense spending is to rise then it’s vital this is managed well, that it develops UK industrial capacity, and that wherever possible dual-use is encouraged, to ensure the maximum spin-offs to the rest of the economy. But here too capacity is limited. Qinetiq which could once have helped with this role was privatised and lost whatever strategic capability it once had. There have been some successes with spin-outs linked to GCHQ. But the fundamental question is whether the close alliance with the US, which underpins so much of UK defense strategy, remains viable if the Trump/Vance era marks an irreversible shift rather than an aberration.
8.Global Policy and advocating for new multilateral institutions
The parallel global issue to all of this is how a progressive government should connect its science policies to global politics. Many of the world’s biggest problems require multilateral answers, and most of them involve science, whether climate or plastics, quantum or space.
But so far the government has given little hint of what it thinks. The moves to support AI Safety Institutes, initiated by Rishi Sunak, were well-intentioned. But they have had little real-world impact yet, and the focus on safety always looked like a misreading of the breadth of governance issues AI was throwing up, and, worse, may have impeded more ambitious ideas like creating an IPCC for AI (and there is no doubt that some in the industry deliberately promoted the focus on safety as a distraction to further delay any regulation).
A Labour government in the mid-2020s should be promoting an energetic multilateral agenda around science and diplomacy, even if the US is pushing in a different direction. The work on an IPCC for chemicals and plastics is a good example. At some point, the world will have to act and there are advantages in being ahead of the game. But there has been little sign of this yet in Labour’s foreign policy.
9. Evidence as an input to policy not as a definitive guide
DSIT should be an advocate of evidence and facts – and confident in asserting these against populists and conspiracy theorists. I’ve long been an advocate of using evidence as much as possible to guide policy and was involved in setting up several what works centres, running various observatories and when in government ensuring that evidence reviews preceded any policy design.
But the ‘more-ists’ consistently get evidence a bit wrong. They assume that evidence alone can guide policy. Instead, all the ‘evidence on evidence’ shows a much more complex picture. Many different types of knowledge have to be synthesised to make decisions – economic, political, psychological, implementation and more (I listed at least 20 types of knowledge that were relevant to decisions on COVID lockdowns for example). Moreover, there are very few fields where the evidence is comprehensive enough or reliable enough to be a sufficient guide to action.
Evidence synthesis is a useful, indeed vital input. But it is very different from synthesis for action which is what officials and ministers have to do in the real world.
This simple distinction is lost on many of the key decision-makers (although most people who navigate the boundaries of research and policy understand it well – and I attempted a summary of the UK experience here). Huge sums are currently being invested by the Wellcome Trust and others on evidence synthesis around the world but apparently untouched by the practical experience, and indeed the science of evidence. The risk is that this will, once again, produce large repositories of evidence that simply go unused. My sense is that this is another symptom of the unhealthy dominance of ‘more-ism’ and the lack of proper debate about its assumptions.
10.Truth – and fighting for it against its enemies
Finally, and perhaps the most important issue of all for a Labour government sometimes at war with Elon Musk’s X, is the status of truth. It’s fashionable to say that there are now only multiple truths – and so no point in trying to regulate the platforms or fight too hard against disinformation. But an alternative view sees this is a battleground worth fighting on, indeed essential if we are not to see our democracy steadily undermined. In the longer run, we will need legal rights to truth (and penalties for powerful organisations that knowingly spread lies), which can build on laws in finance, consumer advertising, and the processes of science. As George Orwell put it, the essence of free speech is the right to say that two plus two equals four, not the right to say that two plus two equals five.
My guess is that this will become obvious. In the meantime, a first step would be to create new institutions to protect elections from AI and deepfakes – vital steps that are needed before the next election. Since science is founded on the continuous search for truths, this should be relevant to DSIT; and since disinformation is probably the biggest single threat to Labour, this should be territory on which Labour is willing to fight. But the silence from ministers has been deafening, presumably for fear of annoying Elon Musk.
Where is the thinking?
So why is there not much sign of thinking or debate about these questions, let alone a more distinctive Labour or social democratic perspective? One obvious reason why the government’s policy appears barely changed from its predecessor is that key roles are now filled by figures from previous Conservative governments.
But the broader reason is that the government remains strongly influenced, some would say captured, by these two powerful London-based elites. As I’ve shown the bio-medical elite controls most of the key jobs and has long had a powerful grip on funding and policy. Although they are highly competent within their field, and have contributed to UK excellence in life sciences, they have serious blind-spots on much else – including all of the issues listed above, which are crucial ones where science meets politics and policy.
The other elite is linked into Silicon Valley and US AI and includes Microsoft, Google, Meta, Oracle and others. Some of the key individuals, such as Demis Hassabis, are hugely impressive. But this elite too has glaring blind-spots and many of its instincts have become increasingly problematic, from blocking any regulation of AI to tying the UK closely to an increasingly politicised US tech-industrial complex which Joe Biden warned about in his last speech as President.
Both elites systematically ignore fundamental political questions – which are questions about who benefits and about values – instead favouring incumbent interests with access and blocking out healthy democratic debate.
The public are sceptical of ‘more-ism’
One factor that should concern politicians is public opinion, which is at odds with crude more-ism. Nesta has published detailed research on UK public attitudes to science and innovation which show that there is a significant minority, perhaps 10%, who essentially support ‘more-ism’: they tend to be male, highly educated, and high income, and concentrated in cities like London. They also include much of the decision-making elite.
At the opposite end of the spectrum are ‘less-ists’, or Luddites who are inherently sceptical or hostile. However, much larger groups have more nuanced views. They are generally favourable to science and technology but worried about effects on the environment, waste, potential harms to children, questions of ethics, and who actually benefits. This should also be the rough reference point for government policy in a democracy, especially a centre-left government. But the ‘more-ists’ are sometimes tone-deaf when it comes to issues of equity, public participation or ethics.
Integrated thinking
As ever more government policy involves science, it’s becoming vital to cultivate integrated thinking, which simultaneously grapples with the science and the politics.
Historians point out that recent science policy documents are strikingly less sophisticated than their equivalents 20, 40 or 60 years ago: an odd mix of cleverness and dumbing down. The dominance of a fairly small group of (mainly men) who appoint each other to jobs and reviews has blocked out precisely the kind of vigorous argument, informed by evidence, which drives science and can also drive better science policy. There is little space for the scepticism that is the motto of the Royal Society (‘nullius in verba’); little space for social science let alone humanities; and little space for intelligent argument.
I’m sure Peter Kyle will go far. He has all the qualities of an effective politician and minister. But he now needs a reset: to shift gear, gently extricate himself from the science elites and remember that he is a democratically elected politician, there to serve the public not vested interests. That reset needs to mirror the reset needed in other fields where Labour often seems to have forgotten its core values and identity.
Just arguing for more doesn’t cut it now, whether politically, intellectually or as a programme for government. It wouldn’t in normal times. But it is even less convincing in a time of profound geopolitical turmoil and a time when the government’s relationship with the public looks increasingly frayed.
Geoff Mulgan is a professor in UCL’s engineering department; a former director of the Government’s Strategy Unit and No 10 Policy Unit; chair of a recent EU programme on ‘Whole of Government Innovation’; and author of ‘When Science Meets Power’, published by Polity Press in 2024
The points about science policy going backwards and becoming less informed by the social sciences & humanities are really interesting. Suggests that excessive deference to scientists & technologists has become entrenched in UK policy.
Makes me wonder if there's an upstream problem about education/curriculum. Feels like the UK system establishes quite a hard & early divide between science & humanities which could then lead to a naive/crude 'scientism' i.e. "I'm from the humanities, I better not interfere with or question the science boffins"
Perhaps a more holistic approach would produce politicians/elites more able to question or critique science & technology in a nuanced way?
Perhaps this is only obvious in retrospect, but it's hard to see DSIT being anything other than a more-ist institution. Primary responsibility for many of the other important dimensions of a complete science and innovation policy sit with other departments (e.g. adoption is with Business, regional growth is HCLG etc).
The original idea might have been for DSIT to coordinate, or possibly even direct, across government, but UK government is pretty terrible at that even in the most propitious circumstances, and DSIT has few levers to pull (it even gives away most of its budget, to UKRI!) There's not much left for DSIT other than more-ism, indeed maybe this was why it was created in the first place.