4 December 2019

The Global AI Index

AI in the UK

Britain is in bronze position in our global ranking of countries. But maybe not for long

By Alex Clark, Ella Hollowood, Imogen Harper and Alexi Mostrous

The UK risks slipping behind in the great global AI race thanks to a failure to transform brilliant academic potential into world-beating performance.

This country currently lies in third place – behind only the US and China – according to a pioneering assessment of artificial intelligence capability.

On Tuesday, Tortoise Intelligence unveiled its Global AI Index, which ranks 54 countries against more than 150 indicators, including quality of AI research, presence on coding platforms, level of investment and government support.

The Index reveals how AI adoption is accelerating around the world, with more than 50 countries announcing national strategies on the technology in the last two years. AI implementation is likely to have significant effects on how societies function thanks to its ability to reshape industries from healthcare to defence.

While the US and China are “streets ahead” in terms of AI capability, Britain is currently leading the second tier thanks to superb home-grown researchers, a healthy start-up scene and a history of innovation in computing, our Index found.

Yet the country is hamstrung by one of the least efficient patent regimes of any major country, the data shows. Only five per cent of AI-related patents filed in the UK have been granted, compared to 33 per cent of those filed in the US. And of those granted, the UK’s patent office took an average of 1,330 days for sign-off, compared to a global 905-day average. In China, the fastest growing AI nation, patent approval takes just 504 days – almost three times shorter than in Britain.

The UK’s AI industry is also claiming a tiny share of the world’s patent filings, suggesting that promising inventions often get stuck at a proof-of-concept phase. Only 0.6 per cent of the world’s AI-related patents have been filed by a British applicant and British-based inventors have filed only 6 per cent of AI patents in the UK itself, according to our analysis of Google’s public patent data. By contrast, American firms are behind 16 per cent of the AI-related patents filed in the UK. Two thirds of US AI patents are filed by Americans.

“The UK has great natural assets – talent, research pedigree and imagination,” Tim Gordon, a partner at Best Practice AI, told us. “But UK industry is still failing to adopt AI with much seriousness, with most applications being proof-of-concept point solutions rather than the deeper strategic change. It’s spectacularly weak at getting AI patents. We don’t have in-depth industrial take-up, nor a positive vision to a build broader consensus for action. We could ‘win’ but, on the current track, probably won’t.”

Britain boasts the highest number of top-rated centres for AI research of any nation – from Sheffield University to the Alan Turing Institute, the index shows. Its academics have published more than 1,100 AI papers in top journals over the last two years, beating Australia on 752, Singapore on 468, and Germany on 276.

However, AI graduates are often tempted abroad by higher salaries or greater investment opportunities offered by foreign competitors. Last year more than 60 British PhD students moved to China alone. The UK is on net zero for talent: we’re losing as many of our own AI PhD students as we’re gaining from other countries.

Filip Kozera, 23, studied informational engineering at Cambridge but chose the US to launch his AI start up, Kristalic.

“Straight after uni I flew to San Francisco with $3,000 in my account and with big dreams,” he told us. “I wanted to start a company and in Silicon Valley, it’s easiest to find funding for an early-stage company.”

Kozera said that stringent US visa requirements often stopped UK graduates moving to the US, but added: “In the UK a good graduate who’s just finished informational engineering could get around £45,000. In the US the starting is like $100,000-$125,000.”

A recent pay survey in the US found that AI engineers at Uber received average yearly salaries of £244,468, while Netflix handed out $206,444. In the UK the average salary for a machine learning engineer is £67,000, according to Reed.co.uk.

One AI area where Britain is leading the way is in healthcare. DeepMind, the British artificial intelligence company bought by Google for $500m for 2014, has built algorithms designed to detect eye disease and plan radiotherapy treatment better than any human doctor.

The country is also home to hugely successful enterprises such as Babylon Health, which connects patients to doctors via a smartphone app, and Darktrace, an AI-driven cybersecurity firm.

Ministers are eager to facilitate the next generation of AI companies. Earlier this year, they announced a nationwide programme of industry-funded AI masters courses at UK universities, paid for by firms such as QuantumBlack and BAE systems. Computer science is Britain’s second fastest growing degree. It has been estimated that AI could add an additional £630 billion to the UK economy by 2035.

However, the Index shows that Britain’s closest rivals are catching up. Canada currently beats Britain on three of our seven “pillars” of AI capability: talent, development and government backing. Talent includes the number of data scientists living in a country; development primarily measures patent filings; and government backing collates all public investment in AI. In Canada’s case, it was the first to launch a ministry for AI, with Montreal and Toronto boasting high concentrations of AI researchers and start-ups respectively.

Germany is closest to catching the UK in terms of infrastructure and development, and beats it in terms of government backing, the Index finds. Last year, Chancellor Merkel announced she would invest £3 billion in AI in an effort to catch the US and China: one of the larger sums earmarked by a government.

France, too, is a key player with particular talent in research and creating a welcome operating environment for AI firms. In 2018, France announced an alliance with Germany to work together on AI research.

Although Britain has announced similar initiatives, the UK does not have a single department dedicated to the growth and regulation of AI. Sources told us that senior figures at the Government Digital Service, the Office for Artificial Intelligence, the Information Commissioner’s Office and Ofcom were fighting amongst themselves for control.

The success of other countries in attracting top AI talent speaks to what it perhaps one of the biggest threats to the UK’s third place in the Index: Brexit and a tightening of migration rules. Two fifths of the UK’s academic workforce in science, technology and engineering are overseas nationals, according to a Royal Society report from July, while research commissioned by the Russell Group estimated that introducing new immigration rules for European Economic Area nationals from 2021 could increase university employer costs by 36 per cent by the end of 2022.

With France and Germany’s AI industries still benefiting from freedom of movement, and with Canada possessing one of the fastest visa processing times in the Index, a post-Brexit UK may find itself usurped.

“Canada is 6-12 months ahead of the UK in terms of working practices and development,” Gordon said. “Whether this is sustainable – especially if a more benign US Government adopts a different approach to immigration/ethics – we have yet to see.”

Dr Kieron O’Hara, associate professor at the University of Southampton told a recent government review: “Other countries and international companies are investing heavily in AI development, but the UK is still regarded as a centre of expertise in research and application of AI, for the present at least. The UK can continue to build on the legacy of Alan Turing and those who have followed him, to remain one of the great centres for AI.”