The UK needs to build a British data library to become an AI leader
Politicians should back calls to develop a British data library if they want to see the UK give birth to the next crop of world-leading AI firms, Jess Jones writes
British startup Wayve has this week received the biggest amount of funding that a UK artificial intelligence (AI) company has ever raised.
Undoubtedly, Wayve is a UK success story but its £800m hunk of cash from Softbank, Nvidia and Microsoft would have been impossible without the vast amounts of data that has enabled its AI technology to transform an everyday car into an self-driving vehicle.
There could, however, be more startups like Wayve if the government embraces a proposal to build a ‘British Library’ for data, put forward by think tank Onward.
Onward is pushing for a secure platform of anonymised data from the UK’s public services. It says this will allow startups, businesses and researchers to easily access the data needed to build AI models, consequently attracting talent and investment to Britain.
The UK’s existing data pool is already the world’s third-largest, according to Tufts University, but this data often fails to match the quality needed for AI use, being inconsistent, outdated, hard to find, or inaccessible.
A similar initiative is already in its pilot stage in the US, where America’s National Artificial Intelligence Research Resource (NAIRR) allows AI researchers to access data from partnered government bodies.
And, in the EU, the Common European Data Spaces scheme, currently in development, will allow data from across the EU to be made available and exchanged in a trustworthy and secure manner, in line with existing data rules.
Onward is right in arguing that the UK risks falling behind in AI research and innovation without its own initiative to unlock data for AI use.
And since the British government’s common refrain is that it wants to become an AI superpower, it should seriously examine the idea.
Fortunately, it is listening. The proposal has already gained the backing from a number of business groups and MPs, including the former digital minister Matt Warman.
When asked about the proposal, Tech Secretary of State Michelle Donelan told City PM: “We’re open to all ideas but we’re certainly not naive to the importance of data and recognise how it is vitally a game changer.
“We are doing somewhat of a data drive within my department and across government to make sure that we are fully utilising data as a key to unlock potential. Data really is the key.”
City PM has asked Labour whether it supports the idea.
But it would also be naive to think that a data library is without its issues, even with checks and balances in place.
Onward has proposed that the library could initially focus on NHS data before expanding to include data about taxation, benefits, crime, immigration, emissions and household utilities.
It is of course vital that any public data like this, even if anonymised, is guarded with the highest level of protection, particularly as cyber attacks are on the rise.
Not only was the actual British Library itself hit by a cyber attack last year, a number of other state bodies have also been engulfed in cyber crises. A “large volume” of patients’ NHS data was recently plastered over the dark web, while the Ministry of Defence launched a crisis response this week after discovering a suspected state-sponsored hack of thousands of personnel in the armed forces.
The National Cyber Security Council has warned that cyber attacks are on the rise due to AI, which is making it easier for hackers to potentially gain access to valuable private data.
But, equally, AI can also be used to help defend against more advanced attacks. For example, Darktrace uses AI to detect cyber threats before they arise. Giving more AI firms access to a potential vast pool of data could actually help offer protection, as well as boost innovation more broadly.
Sadly, the UK is losing companies like Darktrace to American private equity firms and the allure of bigger, deeper-pocketed US markets.
The UK needs more companies like Darktrace and Wayve. A British library for data would help build them.