Interview: Gillian Docherty of The Data Lab: Scotland can rival Silicon Valley as a destination for data science

Scotland has a number of inbuilt advantages, but skills remain an issue

Gillian Docherty is CEO of The Data Lab, an innovation centre set up in 2014 and funded through the Scottish Funding Council with the aim of promoting the use of cutting-edge analytics and data science in the public and private sectors in Scotland through collaborative ventures. Backed to the tune of £11.3m, The Data Lab has goals of adding £104.5m to the economy and creating 250 high-value jobs in the process.

After completing a degree in Computer Science at Glasgow University, Docherty worked for 22 years in various roles at IBM before leaving to head up The Data Lab in 2015. She has given a TED talk and won a number of accolades and awards, the latest being made an honorary Doctor of Technology (DTech) by Aberdeen's Robert Gordon University this month. She believes Scotland can be a leading destination for data science to rival Silicon Valley or London.

Explainable algoritms and big data ethics

One of the ways in which The Data Lab seeks to boost Scotland's digital economy is through supporting startups - it runs competitions and hackathons and organises summits among other things. Computing asked what area she would focus on if she were starting a tech company today.

It's not good enough for the bank to say ‘computer says no'

"There's a big gap around explainable algorithms and ethics," she said. "So it would be something around what tools or capabilities we can build such that algorithms can be explainable rather than a black box. And I'd start with a regulated industry like financial services. If you're turned down for a loan it's not good enough for the bank to say ‘computer says no', you need to know why."

The need to understand the often mysterious ways that neural networks arrive at the decisions that they do has been made more urgent by the recent shenanigans involving Facebook and electoral targeting.

"The spotlight now is on the ethical dimensions around the use of data and algorithms and AI. Just because we can, should we? I think you'll see more work and more startups coming into that space," Docherty said.

A Goldilocks economy

The creation of ethical frameworks and regulations around AI will ultimately require global collaboration (not so easy in these fractious times perhaps). However, most data science and machine learning efforts are local and use-case specific. Scotland is home to a high concentration of data-intensive sectors such as financial services and energy, which is a bonus when it comes to attracting talent, she said.

Aberdeen, home to the oil and gas industry

"Scotland is a Goldilocks economy for technology, neither too big nor too small. It's big enough to attract companies here but small enough to be really connected. There's a really vibrant tech scene."

She continued: "We've worked extensively to make sure businesses move here, build their data teams here and extend their reach here. It's a very vibrant and supportive community only an hour's flight from London and I think that because we connected like that it does set us apart a little bit. We'll never compete with London or Silicon Valley in terms of numbers, but actually we‘ve got something a bit different and we've seen that attract companies to build their teams here."

Other factors in Scotland's favour include relatively low house prices and wages and the presence of a number of top universities.

Despite all that, though, for businesses finding the right skills locally can be tough.

"One of the biggest challenges is around skills and talent - and that's at all levels," said Docherty. "We've had to significantly expand our programme and add to our capability to at least start to address that issue, but we're still some way from solving the problem."

Addressing the shortfall in data science talent requires a concerted effort to boost tech skills across the board. That means courses for CEOs to teach them how to make the best use of their data, new PhDs with a strong vocational component, MScs to attract the best overseas talent, apprenticeships and continuous personal development (CPD) via onsite training and online courses.

All of these things need to happen simultaneously if the country is to create a robust base from which to grow, and that means building on existing strengths.

Last week The Data Lab launched its third domain-specific 'massive open online course' (MOOC) for data science. A collaboration with Dundee University, this one is focused on gaming. (The other two are in medicine and social care.) There are plenty of "data science 101-type courses out there", said Docherty, but few that are focused on the needs of specific industries.

"The university led the effort with support from five gaming companies in and around Dundee, so they're shaping the curriculum and potentially delivering some of that too. That should make it very relevant to those working in the games industry," she said.

The Data Lab also supports a number of data science MSc courses. The focus here is on widening access to include students who might not have a first degree in a STEM subject.

"We have students going through our data science MSc programme from sports science, medicine and even law. You need to put in the effort to make sure you are attracting that diverse background because it's not all computer scientists and statisticians," she said.

"In our cancer challenge project [an open innovation venture with the goal of using analytics to improve cancer care], we worked with medics who are also part-time coders and when you combine that domain expertise with an interest in the technology side you can see some amazing results."

Priming the talent pipeline

Combining sector-specific experience with advanced technical education is the goal of the new PhD courses introduced by The Data Lab, in which postgraduates spend 80 per cent of their time working in organisations and businesses on real data problems.

"We piloted that with St Andrews University and it proved really popular with industry. The pace of change is such that many of the industry partners that we work with don't necessarily have the time or the capability to focus a PhD on very, very narrow topic, so the engineering doctorate programme sees the students awarded a qualification on the back of five- or six-monthly projects and will hopefully provide the industry partner more benefit than the classic PhD, which we do as well. We think there's a need for both."

Docherty aims to make Scotland a destination for data science

However, even with the increased number of graduates, postgrads and apprentices flowing through the educational pipeline there is still not enough talent to fill all the gaps.

Relying on the education system to supply talent is not going to be sufficient

"We have a need to retrain and cross-skill existing workforces with new technologies and capabilities because relying on the education system to supply talent is not going to be sufficient. There are opportunities for those already working with data but not quite understanding the new techniques and capabilities. We need to ramp up the availability and accessibility of CPD."

Many of the projects The Data Lab is involved with are pan European. Has Docherty seen any impact from the Brexit machinations? The main one, she said, is uncertainty about whether the number of foreign students coming to Scotland to study data science will be hit. "How can we make sure we are as welcoming as we can be and create the opportunities to attract them?"

It's possible too that they'll be a cooling of collaborative international efforts in areas such as AI and cybersecurity, she thinks.

Thus far though, The Data Lab seems to be on track to achieve its goals. It has invested £2.5m and its partnerships have achieved revenue gains of around £80m with the creation of 280 jobs.

Computing's AI & Machine Learning Live! summit is happening on November 19th. Register today.