A measured approach to funding

Is long-term funding a better model of financing science research than the never-ending cycle of grant-applications? NPL think so – and they are putting their money where their mouth is when it comes to their new AI and Big Data fellowships…

The UK economy is now undergoing a transformation similar in scale to previous Industrial Revolutions that gave us mechanisation and electrical power. The present-day economic revolution will see society rebuilt around mining data to power products, systems and services just as our economy was once redesigned around mining energy to fuel machinery. As The Economist recently argued “the world’s most valuable resource is no longer oil, but data.”

The UK is moving to take advantage of this trend; the government’s recent Digital Industrial Strategy pledges to help “every British business become a digital business” and the landmark UK Industrial Strategy, which pinpointed AI and data as one of our four Grand Challenges, explicitly aims to put the UK “at the forefront of the artificial intelligence and data revolution”. The government is also pumping £0.95 billion into a Sector Deal for AI and Big Data and creating a new ‘AI Council’ to help industry and academia join forces to unlock the commercial and social potential of Big Data.

Offering scientists the potential for job security is critical to attracting a top tier of scientists with large-scale, long-term research ambitions

For its part, business is investing heavily in the necessary technological overhaul needed to realise this vision; the private sector has matched a government investment of £300 million in UK AI capabilities. Initial projects include a pioneering research partnership between Rolls-Royce, the Alan Turing Institute and Cambridge University launching a new AI supercomputer which will be made available to industry. A third of more than 2,000 industrial companies surveyed have already started to digitise their entire supply chain processes and nearly 75% will have passed this milestone by 2020.

Big project, big money As academia and industry join forces to drive the revolution, the potential benefits are enormous, with AI projected to add £630 billion to our economy by 2035, and Big Data estimated to be worth £322 billion to the UK by 2020. By increasing consumer spending through greater personalisation and diversification of products, AI will also boost productivity by 14.3% and GDP by 10.3% in 12 years.

Yet the success of this entire endeavour hinges on the extent to which we can trust the validity of the data fuelling the new economy. Without data confidence, organisations will not be able to use digital information effectively or make confident decisions based on it. Crucially, as AI systems are increasingly entrusted with safety-critical tasks, from diagnosing diseases to navigating vehicles, we need absolute confidence in the algorithms making these decisions and the data they draw upon.

Big Data is defined by the ‘seven Vs’ (Volume, Velocity, Variety, Viscosity, Variability, Veracity and Volatility) and ‘veracity’, or accuracy, is the key criterion upon which all the others depend. In order to be world leaders in AI and data, the UK needs to become world leaders in data science. We must be at the forefront of efforts to conceive universal gold standards for the consistent and reproducible collection, curation and analysis of data.

This will require us to act now to make the UK a global hub of data science, and research institutions will have to offer new incentives to attract and retain top talent. It will necessitate investment in long-term research projects that ensure world-leading researchers remain here long enough to carry out cutting-edge research with genuine impact and longevity.

If we are to achieve this, we must move away from the traditional narrow focus on short-term headline-grabbing research projects towards a long-term model of research funding that gives scientists the freedom to pursue research beyond normal funding cycles. This would give UK-based researchers the opportunity to carve out their own distinctive niche in the field and create entire new fields of data science that will put the UK in the vanguard of the AI and data revolution. We need to offer scientists the potential for a full salary and research expenses over the entire lifetime of their research.

A long shot For more than a century, NPL has developed and maintained the nation’s primary measurement standards. We are ideally positioned to move from the physical and chemical measurements underpinning the physical economy to the measurements underpinning the data economy.

NPL provided the setting for the development of the ‘packet-switching’ technology by which data is broken up and reassembled on arrival, the technological foundation for worldwide data communication, the internet and the modern data economy. Today, we are aiming to lead the next data revolution by investing in long-term research with the potential to tackle data science challenges affecting our key growth industries, from data analytics to AI. For industries such as life sciences, challenges include unlocking the potential cures hidden in vast quantities of drug development data and how to use AI to speed up disease diagnosis and personalise treatment plans.

Unlike traditional fellowships built around short funding cycles, NPL’s new Measurement Fellowships will offer leading data scientists from around the world the opportunity to receive research expenses and a salary for the entire lifetime of their research project. For research that shows promise, NPL will invest in the long-term and offer the scientist the opportunity to become long-term employees. Offering scientists the potential for job security is critical to attracting a top tier of scientists with large-scale, long-term research ambitions. The Fellowships also aim to attract talent from around the world by offering scientists full access to facilities at one of the world’s leading centres of measurement science.

We must move away from the traditional narrow focus on short-term headline-grabbing research projects towards a long-term model of research funding that gives scientists the freedom to pursue research beyond normal funding cycles

The Fellowships will target international leaders in their field to lead an exciting research programme and help PhD scientists and engineers to develop their research. They will also have the opportunity to work on and bid for a wide portfolio of research, grants and commercially-funded projects. The Fellowships will focus on key data science challenges identified by independent expert advisory groups within the UK industry sectors earmarked for greatest growth by the Government’s recent Industrial Strategy, ensuring the research is aligned with industry needs and generates maximum economic value.

This will appeal to scientists because it gives them the opportunity to leave a lasting real-world legacy and improve quality of life and economic growth in the UK by tackling some of the major social and economic challenges of our time, from energy and healthcare to the environment.

Crucially, the NPL Measurement Fellowships will focus on validation and reproducibility, giving the UK the potential to shape the standards governing international data transactions and become the global leader in providing ‘quality-assurance’ for the data economy.

Good for science, good for the country Shaping the international frameworks, methodologies, analyses and protocols governing the digital economy would put the UK in the driving seat as AI and Big Data expand across the world market. This will be vital for everything from trade to consumer confidence. Just as international trade in physical products requires mutually-recognised quality tests to ensure products are produced, transported and stored according to agreed standards, we need an international framework guaranteeing the integrity of data.

For example, in order to buy data-driven products such as wearable fitness aids, consumers need confidence in the health data they use and the algorithms that analyse it. Better data science is also a matter of life or death.

In order to guarantee medical AI can make a trustworthy, accurate diagnosis, we need traceable standards for how patient data must be collected, stored, tagged and analysed. Critically, we need a digital ‘chain of custody’ tracing data back to its original source to guarantee that all data has been created according to the correct criteria.

Equally, when vital products, processes and systems go wrong, we also need confidence in our ability to take corrective action and understand the causes; if a driverless car crashes, we need a trustworthy digital audit trail to tell us what caused it.

The NPL Measurement Fellowships offering the opportunity for researchers to conduct long-term data science will be the first of a raft of Measurement Fellowships in other scientific disciplines, based on the same model of investing in the future by turning research into a career, not a short-term project. The new model of fellowships offers an example for how Britain can transform the nature of scientific research to position itself as a global leader in the industries of the future.

By supporting long-term data science research designed to address today’s most pressing industry needs, we can give talented researchers the chance to create a lasting economic and social legacy for the UK.

Find out more about NPL’s Measurement Fellowships Programme at fellowships.npl.co.uk

 

Author:

 Dr Jan-Theodoor Janssen is Director of Research at the National Physical Laboratory (NPL)