Christine Foster, the managing director for innovation at the Alan Turing Institute, the UK’s national institute for data science and artificial intelligence. The institute was set up in 2015 with the goal of advancing fundamental research and applying that research to real-world use cases.
In addition, it also aims to train the next generation of data scientists and data science leaders and advise policy-makers and the public on how to adopt this technology responsibly.
In Foster’s view, the institute exists to give the UK a competitive advantage on the world stage of data science and artificial intelligence. But to do so UK businesses will have to engage with the institute to get that advantage.
Organisations can do so by joining a Data Study Group, to which they can bring in datasets and questions they want to find out from those datasets. The Turing Institute will then put on a week-long hackathon to get those answers.
Large enterprises can also fund research and therefore invest in artificial intelligence and data science alongside the UK government. There is also a research engineering group, which Foster describes as “engineering research and researching engineering." She added: "They are the ones who take something out of a paper and put it into the first usable bit of code.”
"Researchers need to know which problems you are grappling with."
It is also possible to join an interest group such as one on data ethics or social data science. “Researchers always have these different research interest areas but they need to know what actual problems you are grappling with in the practitioner community to know that they are working on the right things,” she said.
One piece of research that was incubated at the Alan Turing Institute came about as the result of collaboration between an ethicist, a lawyer and a computer scientist. They decided to work on automated decisions and tackle the problem of ‘the computer says no’ where the individual doesn’t understand why they were given that particular outcome.
“In this case,” said Foster “we are very proud of the effect the policy influence, the regulatory influence has had in pushing counterfactuals as a valid, meaningful way of giving explanations to individuals.” The resulting research paper “sat in this interdisciplinary spot” and gave an early prototype of how to figure out which vector to go along to get from a no to a yes.
This cross-disciplinary approach is reminiscent of Turing’s own career, as he led a multidisciplinary team at Bletchley Park tasked with deciphering German Naval communications in World War II.
He also conducted research in many fields and his last published paper was actually on the subject of chemistry which looked at the mathematical structures of the hormones in animals’ skin that cause the colouring on their coats of spots and stripes.
Foster herself has worked across many fields, having been employed in consulting, the music industry, the payment card industry and payment processing. When she was in the music industry at EMI music, she was working with colleagues to collect data to better understand fans and their interactions with the bands.
“Once I was scraping MySpace data and we had put so many demands on their website that they had thought we were launching a denial of service attack, but we were just trying to get the data we wanted to know what was going on with bands and MySpace,” Foster said.
She explained that they were getting information from radio, sales systems and scraping information from the web.
“We were building the relationship between all these different touchpoints of fans and artists and the eventual outcome of how the album did and how engagement would go, so how profitable might this album be based on everything we knew. Many would recognise that today as mixed modelling,” said Foster.
This would have been between the mid to late 2000s, a time when Foster said this type of data gathering would have been ground-breaking, as the heterogeneity of data sources at the time was a difficult challenge to overcome.
Taking a high-level look at the industry and the skills needed to operate within it, Foster said it is not necessary for everybody to learn to code. She said: “I think it would be a massive overcorrection. I know that we have lots and lots of open roles to fill but I did say the What-If tool is for non-programmers.”
Foster said that advances in technology are making it easier to use and a lot more accessible. “Actually it is making tech bigger and lots of different people can get involved."