Sign up for our newsletter
Policy / Big Tech

The UK’s National AI Strategy shores up its strengths – but who will benefit?

A new National AI Strategy will help co-ordinate the UK's considerable AI capabilities but Big Tech might draw the greatest benefit, critics warn.

The UK this week published its new National AI Strategy. One of the pillars of the country’s digital policy platform, the strategy will help coordinate cross-departmental AI initiatives, experts say, and shore up the UK’s established strengths in the AI arena. But critics warn that the benefits of public investment in AI may accrue to the tech giants.

UK National AI Strategy

Culture secretary Nadine Dorries MP identified Alan Turing as an “AI trailblazer” when unveiling the UK’s new National AI Strategy. (Photo by Lenscap Photography/Shutterstock)

What’s in the UK National AI Strategy?

The UK’s National AI Strategy identifies three overarching objectives – investing in the long-term needs of the AI ecosystem; ensuring that AI benefits all sectors and regions; and governing AI effectively – and a series of short, medium and long-term measures with which to achieve them. These include initiatives to promote AI skills education, coordinate funding for AI research, and developing the UK’s open data and compute infrastructure.

Although it is the country’s first official national AI strategy, the UK has been an early adopter for AI policy. The OECD’s AI Policy Observatory counts ten national AI policy initiatives by the UK to date, second only to the US. These include 2017’s ‘AI Sector Deal’, which contained many elements of a typical national strategy, says Karine Perset, head of the observatory

The UK also punches above its weight in many of the AI performance metrics that the OECD observatory tracks. For example, the number of top-quality AI research papers published by UK researchers in 2021 is projected to be just shy of China’s figure. And in 2020, UK AI companies attracted almost as much venture capital investment as the EU put together.

White papers from our partners

As with other digital policy platforms from the UK government, the National AI Strategy makes economic growth a priority. "With the help of our thriving AI ecosystem and world-leading R&D system, this National AI Strategy will translate the tremendous potential of AI into better growth, prosperity and social benefits for the UK," said Secretary of State for Business, Energy and Industrial Strategy Kwasi Kwarteng on its launch; "AI will be central to how we drive growth and enrich lives," added new Culture secretary Nadine Dorries.

This focus on growth is in keeping with other countries' national strategies, Perset says. "All countries are being very commercial about it, except maybe for Canada, which has a very strong focus the UN Sustainable Development Goals" in its AI strategy.

But there are some features that distinguish the UK's National AI Strategy, Perset observes. One is its aim to develop the country's compute infrastructure of AI applications. "Access to computing power is essential to the development and use of AI, and has been a dominant trend in AI breakthroughs of the past decade," the strategy states. "To better understand the UK’s future AI computing requirements, the Office for AI and UKRI will evaluate the UK’s computing capacity needs to support AI innovation, commercialisation and deployment."

Despite its significance, the only other country to address compute capacity in its AI strategy is the US, Perset says. "There's a lot of focus on algorithms and code, but there's comparatively little in terms of the high-performance computing or cloud computing that is needed to support these ambitions at a national level," she explains.

Another distinguishing feature of the UK approach is its focus on international technical standards as a mechanism for governing AI. While other jurisdictions, most notably the EU, are developing AI legislation, the UK (along with Australia) is more focused on influencing technical standards for AI systems, such as requirements for documentation and reliability testing. One of the new strategy's long-term commitments is to "explore with stakeholders the development of an AI technical standards engagement toolkit to support the AI ecosystem to engage in the global AI standardisation landscape"

"The UK and Australia have been quite prescient," says Perset. "They realise that [international standards bodies are] where a lot of the decisions are going to be made. Whatever standards we develop in those venues will dictate how [AI] systems are developed. And so we need to make sure that the standards reflect our objectives."

Also of note is the commitment to "roll out new visa regimes to attract the world’s best AI talent to the UK". This is another uncommon policy, Perset says, and especially noteworthy as the UK is already a net importer of AI talent, according to the OECD's analysis of LinkedIn postings.

In all, Perset says, having a single National AI Strategy will help the UK capitalise on its strengths and coordinate its activities. "What's very interesting about this strategy is it's evidence based, taking into account the relative advantages of the UK in the AI space, of which there are quite a few," she says. "The advantage of having one strategy is really helping that coordination to take place across government."

Who benefits from national AI strategies?

Not everyone is so enthusiastic. Wim Naudé, professor of economics at University College Cork, has been a vocal critic of national AI strategies, which, he argues, overestimate AI's benefits and understate the drawbacks. "These ambitious national AI strategies are all unashamedly optimistic," he says. "[They say] 'AI is only for the good and we can have all these wonderful applications that will accelerate economic growth'. They never tell us the price [which] is, inevitably, our privacy."

Naudé is sceptical of the economic value of 'AI', in its current meaning of machine learning and other associated techniques. This kind of AI is a "glorified regression technique," he says. "It can find needles in haystacks in big data and it can make predictions. [AI applications] are at the margin useful, but they're not really impactful." He points to a recent survey of 800,000 US companies, of which only 3% are using AI, as evidence of its limited business value.

Instead, it is the tech giants who have the data, skills, infrastructure and business models to benefit most from AI, Naudé argues. A lack of AI skills, for example, is only a problem for big companies with sophisticated AI operations. "A small business is not going to employ three PhDs to work on data sets that they don't have." There is a risk, then, that national strategies to bolster AI capabilities are of primary benefit to the tech giants, he says.

Perset agrees that this is a concern, and one that other countries have addressed in their AI strategies. In Spain, for example, public funding for AI research is reserved for consortia that include a local SME and a public institution. Policies such as these "force technology transfer from these large companies to SMEs and public institutions [which] can't compete with the companies for the best talent", she explains.

Naudé, meanwhile, argues that government resources would be better directed at more fundamental research, such as advancing AI through neuroscience. "We don't understand the human brain. We don't know what consciousness is, what its purpose is, or how it arises. So there's a lot of work in understanding the basic concepts of intelligence that we need to invest in. But we're putting all our money into short-term things."

Pete Swabey

Pete Swabey is editor-in-chief of Tech Monitor.