Interview: Ana Basiri, director of The Centre for Data science & AI at The University of Glasgow
Interview: Ana Basiri, director of The Centre for Data science & AI at The University of Glasgow
After the recent launch of The Centre for Data Science & AI at The University of Glasgow, it’s director, Ana Basiri, spoke to Holyrood on the details behind the development of the site and it’s hopes for the future.
Where did the idea come from?
If you look at the last few years, you will realise that every university and more importantly, the government have started a few initiatives related to data science and AI. Just to name a few, we started having a national AI strategy just two years ago, with the government setting a plan for the UK to become a global AI superpower in the next decade. And almost all universities have responded because it is a very promising area. All policy needs data to back it.
We recognise the potential of the data; we know that AI can revolutionise the industry and we want to respond to that. So we had a conversation with the senior management in the university, and we thought, ‘the university does a really good job in terms of research, but it is a scattered’, and that has got three issues.
One is we don't know what the gap in our research is because there is no unit bringing everybody together. Two, it's not strategic. And three, probably the most important is AI and data science is a great discipline, but we know for a fact that it's a very multidisciplinary sort of landscape. We can't just rely on organic collaboration between people and so we need to facilitate this.
Having this unit can also help us to be recognised so others can come there as their first gate. The centre can be that sort of facilitator, both internally and externally. It is a response to something that we knew was demanded.
What is the process behind selecting who can join the centre?
One thing that is really at the heart of the structure of the project is inclusivity and fairness. I didn't want to continue on the basis of only inviting the people I already know I like to work with. I want to make sure that I work with the best, even if I don't know them.
There are six programs within the centre. First, we open the call for collaboration within the university so we can make a rational decision of the best group of people to join. Externally, we also talk to the National Institute for Data Science and AI, because they're the orchestrator of the whole ecosystem of AI in the UK, to ask them how we can make it as inclusive as possible. They gave us advice on what was best practice and what other universities did that wasn't or was very good so we could replicate that.
How can industry come in?
We have several people within the university that are called ‘impact facilitator’. They work on three main areas with industry. One is companies in or outside the UK which have a very good data set or research question but just don't know how to analyse that, or who is the best person to respond. So, we get them to come to the centre and talk to our researchers. The second option is the other way around. We showcase some of the research that we do so they know the capability that the University of Glasgow has got. And the third part is co-creating something. These three strategies have been very successful because they have resulted in some PhD students being co-funded by industry, it has also led to the creation of upscaling courses, which I am very proud of because AI landscape is changing very fast.
Has there been a lot of interest in getting involved in the project?
Overall AI and data science is relatively attractive but there is a skill issue across all countries. Universities are struggling because the market is very attractive in terms of the salary. So, our students go to industry and work there, as opposed to staying for post grads. But we are working on that, even training some of our staff to take on a slightly different roles within the centre. But to be honest, in terms of recruiting new people, I don't think that's going to be a major problem.
Do you think the ‘constantly changing’ characteristic of the sector is going to impact AI education?
Definitely. Just think of the impact ChatGPT has had in a year. Nowadays, if someone was away for about a year, basically, they know nothing about present AI. I have colleagues that got pregnant, and when they came back to work they basically didn't know anything. It is also not something that you get a degree and you're good to go for the next five years.
I think that constant training education and more importantly interaction is very important – so we created a team where everyone can learn something from each other. By bringing everybody together, people can act based on their own interest and needs and still deliver very good overall research at the end, because the centre facilitates that.
Are there any current studies that you think have the most potential for change?
Well, there are a few projects. For example, there is one focusing on health inequality. In Glasgow there’s a major problem, as between West to East Glasgow – which is a 15-minute train journey – there is a 15-year difference in life expectancy. We are working with Glasgow City Council, health professionals and data scientists to understand what is happening. We looked back at all the policies from a data perspective to then try and inform the policymakers on how to improve the situation.
Another interesting project is one related to cancer research. For 50 years, everyone who was born in Scotland has a unique record of their genomics, and that data has not been explored very well, simply due to ethical issues and because we don't have a trusted research environment that allow us to do that. But this has so much potential, as if we do trustworthy data science we can understand what the impacts are on areas like education. So that's one we are hoping to carry out with Cancer Research UK.
What are most interested in?
One of the things that I'm very interested in personally is the kind of environmental and legal issues of data itself. Most of the time we think data, or digital, is greener. So, if I have this conversation over Teams, it's better than if I drive to an office to have it. But apparently, that's not the case. Data centres also need energy to cool down the servers and everything related to that. There are very good studies showing that if the internet was a country, it would be the fifth most energy-consuming country in the world. We need to talk to the public about the implication that every time I send an email, there is a carbon dioxide emission associated with it.
There are also other implications like when I die, my Google account, my Twitter account is still live and GDPR doesn't protect that because it is a right for living people. And if it is hacked, my family cannot go to police and sue a person because of this. So, we’re work with Scottish Green Party to come up with some sort of motion for October to make sure that there we consider putting digital inheritance legislation in place. Everything on earth dies at some level, either by design or naturally, but digital cannot stay forever.
Will the results be accessible to everyone?
One of the reasons behind the whole idea behind of the centre was most of the time we do the same research because we just don't know that it has happened before. Reproducibility is something that we are really passionate about, whether it is a positive result or not. So, we really try to work with some of the leading institutions like Open Data Institute, to make everything available to everybody. We have recruited some research software engineer and data scientists to make sure that if I do some research regardless of the outcome, they can publish it as a part of the centre output available for everybody.
Again, I just want to go back to the terminology of accessible. Sometimes we do fantastic research that is technically accessible, but people do not understand it. One of the roles of the centre is to train celebrity scientists so they can communicate with the public in a language that they understand. This way we can avoid misinformation about AI in both sides of the spectrum so we can have a realistic conversation as well as understand the public’s concerns. We haven’t begun this yet, but we want to work with, for example, newspapers and journals because they have the expertise on how to translate something into ‘normal language’.
Why is the centre specifically important in Scotland?
One of the important things is in Scotland, just due to geographic reasons, we're relatively remote from many of the things that happens in, let's say, the Golden Triangle of Oxford, Cambridge, London.
Also, in Scotland there’s different data legislation or there are different fundings available and different focus of the research. For example, Scotland has their own AI strategy separate from the UK Government. They both have huge interests that sometimes overlap or are prioritised differently. That’s huge potential because that means something that in England is recognised as important, but they don't have resources to do that, we can take that as our priority and make a slightly better response as a whole nation.
What are the long-term plans of the centre?
We certainly want to expand. Internally, there are some areas of research that are growing, and we want to support them. There are also some new posts that we want to advertise so they can develop areas of research that we don't do much about.
Externally, we have responsible AI investment that is £34m-worth of funding that UKRI, our national funding agency, put in place and Glasgow is a leading part of that.
We want to make sure that we don't duplicate the research, but we also want to make sure that we maximise. That's really a major part of our plans, where we have a lot of administration in place to make sure that our communication remains sustainable with other universities, industry, big technology companies, government, and research councils to make sure that whenever there is a need, we can get involved or lead something.
So yes, expand, evolve and become an effective player in this ecosystem is really the main objective.