On the outskirts of Edinburgh, a climate-controlled facility is quietly homing a row of large boxes, which at first sight look not too dissimilar to soft drinks vending machines. The colourful array, however, dispenses not fizzy juice but instead are powerful computers designed to perform billions of calculations in mere seconds. Affectionately named ‘Archer’, these vast number-crunching machines, once the latest upgrade is complete, will be able to carry out truly unimaginable computational leaps.
The ‘30-petaflop’ system is part of a £79m upgrade to Britain’s supercomputing centre, which sits in a business trading estate outside of the capital. Once it is completed in summer, it will be one of the concrete steps on the way to remaking Edinburgh and its surrounds into the ‘data capital of Europe’, a vision inherent to the data focus of the Edinburgh and South East Scotland City Region Deal. One of the services the upgrade will support is the new Edinburgh International Data Facility (EIDF), and its opening will be part of a timed sequence of unveilings for core elements that support the university’s DataDriven Innovation (DDI) initiative.
Jarmo Eskelinen, director of the programme and former chief innovation and technology officer at the Future Cities catapult in London, says: “The facility is at the heart of the programme and much of the ambition we have for data will depend on the investment into its computational power. It is critical, and underpins the entire programme.” The EIDF is very much the hub for the city’s digital and data ambition; although configured as a high security data centre, many of its services will be delivered on the Cloud. Researchers from academia, industry and Scotland’s public sector will increasingly be able to analyse large data sets to drive the coming wave of automation and artificial intelligence. This has the potential to revolutionise the way we do healthcare, predict weather patterns, test engine safety in aviation, and model financial risk.
It would not be an exaggeration to say that the £660m DDI initiative will be the centrifugal force powering a host of new research facilities that are being built in the coming years. Already, the University of Edinburgh’s Bayes Centre, which opened in October 2018, has disrupted the notion of what constitutes an academic building. The open plan, multi-storeyed facility has become a hive of activity for PhD researchers in diverse data-related research fields to mix with companies and third sector organisations pushing the boundaries of technology: names including BigML, Cognihealth, Jearni, neurolabs and SageCity mingle with The Data Lab – Scotland’s AI and data science innovation centre – and Nesta, the innovation foundation. That co-location and co-working model will be pursued when DDI activities of three other hubs eventually come on stream. In 2021, the National Robotarium at Heriot Watt University is scheduled to open.
The Edinburgh Futures Institute (EFI) – an interdisciplinary hub for social and data science, the arts, and humanities – will occupy the restored Old Royal Infirmary in the city’s Quartermile in 2022. The new HQ of the Usher Institute for people, populations and their data will open in 2023 at Edinburgh Bioquarter, near the city’s main hospital in Little France There is a wider play to all of this strategic investment, of course. Outside of London, Edinburgh’s tech sector is well regarded and the city’s CodeBase is the UK’s largest technology incubator, a thriving mix of coders, entrepreneurs and socially-minded technologists. The DDI programme, like many government-backed initiatives, has a number of social and economic goals at its core, including delivering ‘inclusive growth’, which aims to positively contribute to longterm societal goals. Using public sector data more productively, and equitably, has the potential to reshape healthcare services, for example. The ‘NHS Data Loch’ project pools the electronic health records of the East Lothian region in a machine learning environment and makes it possible to ask new questions from old data.
Last year, researchers used data of over 40,000 patients to study sex-specific laboratory thresholds for heart disease and revealed fives times more additional women than men with myocardial injury. AI systems accessing health data can also be used for virtual clinical trials which might be able to simulate the outcomes of a therapy, making it possible to avoid costly real-world trials doomed to fail. Crucially, there’s potential for considerable commercial benefits f new solutions can be developed and tested both faster and more precisely. “This is all about how we can embed innovation in real-life practice in the region,” Eskelinen says. “One might say that universities have not always been so good at looking at place-based commercialisation opportunities, but with this deal we have an obligation to make sure that we deliver scalable results to our region. We are used to providing safe access to these datasets for researchers without jeopardising security or privacy, so we have these nuts and bolts of building a well-managed operation, which can be delivered at speed using the cloud.” Every global city is looking for its own ‘edge’ in the data world.
It can be said that the social goals within the programme provide that advantage, as they link to the progressive agenda of government. But there are other fields within data that are yet to be cracked. Scotland is embarking on a national Artificial Intelligence strategy, which will be published later this year. Part of that process will be to engage with the challenge of devising a regulatory landscape that is fit for purpose when it comes to delegating human decisions to machines. Shannon Vallor, a former consultant on ethics for Google and professor at the Santa Clara University in California, arrived at the University of Edinburgh in February as the first Baillie Gifford Chair in the Ethics of Data and Artificial Intelligence (AI) at the Edinburgh Futures Institute (EFI). Her work will help the university – and inform the national strategy through her role as the Chair of the Scottish Government’s Data Delivery Group – as it rises to the challenge of developing a working model to mitigate any biases that could impact the way AI solutions are designed.
Eskelinen believes this is all vital to ensure that Edinburgh ‘does data right’, a guiding principle enshrined at the heart of the programme. He says: “Laws and policies are needed but have been slow to develop – there is a need to get our policy frameworks in shape so we can act quicker to make use of data in a way that is compatible with the rights of individuals and organisations. So, we want to study questions or privacy and ownership and to understand how the fair use principle can contribute to the delivery of better public services, allowing all our citizens and companies to benefit.”