Data Summit at #DataFest19: Conference roundup
By Rachel Crookes, Head of SAGE Campus
As part of Scotland’s annual DataFest, the 2019 Data Summit conference took place in Edinburgh on 21-22nd March and was packed full of eye-opening sessions from speakers at the cutting edge of data driven innovation.
I wasn't sure what to expect from the Data Summit. The agenda was pretty mysterious. Was it going to be a total tech-fest? Would the emphasis be on industry, society or something else? Would there be many academics there? And how accessible would it be for a social scientist? What we got was accessible, slick, thought-provoking, well-curated and impeccably hosted by TeenTech founder and ex-Tomorrow's World legend Maggie Philbin. Here are my highlights:-
It all kicked off with Christopher Wylie's blistering and hugely entertaining take on Cambridge Analytica and his role in breaking the story. It was meant to be an interview but Philbin only got to ask two questions, such was the speed and intensity of the story that erupted from Wylie's lips. Breathlessly vital, Wylie ended with a plea for us to think of technology not as a service but as an architecture, and a dangerous one that exists and functions without safety nets or regulation standards. “And what job can you get where you refuse to use the internet? It's not a choice." Many of us thought we knew the CA story, but in the re-telling by Wylie, we realized there was tons more to say.
Dr Sue Black (OBE)
As a long-time fan of Dr Sue Black (OBE), her session was always going to be a highlight and she did not disappoint. Professor of Computer Science, technology evangelist and founder of TechMums, I wonder if she ever tires of telling her story of going back to college as a mum of three, carving out an incredible career in computer science, and inadvertently starting a successful campaign to save Bletchley Park? I would never tire of hearing her tell it. Her next mission? To increase the number of female students doing computer science from 15% to 50%.
“As consumers of data we need to ask four questions: Who was asked? What was asked? How was it interpreted? Why do we care?”
Liberty Vittert (Associate Editor of Harvard Data Science Review) kicked off Day 2 by taking on fake news. Or more accurately, misleading statistics. Vittert was amusing and slick as she unpicked how numbers are accidentally and willfully misinterpreted and misrepresented in the media (alongside how much she loves french fries). From confusing correlation with causation to using numbers that are too big and thus inflate the sense of risk that come with them, Vittert circled back to this fundamental point: always ask the right questions of the right data, because if you don't, it can have devastating consequences and waste a hell of a lot of money. As consumers of data we need to ask four questions: Who was asked? What was asked? How was it interpreted? Why do we care?
Stephanie Hare
Researcher and Broadcaster Stephanie Hare stepped things up a gear with her deeply uncomfortable and challenging session on biometrics. Hare boldly suggested that by the end of her session we'd 'be freaking out', and she wasn't wrong. Hare took us on an exposing and difficult tour of biometrics in action, offering a glimpse into facial recognition pilots happening in the UK and wider schemes in China. She showed some slides without any comment at all, leaving us to silently digest what was on them, and it's hard to explain the palpable impact this had on the atmosphere. In her own words, "If you turn off your brain then biometrics are tempting!" Her call to arms is for better (some!) regulation on the use of biometrics, and her case is compelling.
I needed cheering up by this point and that job fell to Daniel Susskind who was here to talk about the impact AI will have on our workforces and future employability. Which jobs won’t exist in 20 years' time? And what can we do about it now? Spoiler alert: it's not that simple. Susskind's take was that people don't really perform ‘jobs’ at all; instead we perform lots of different tasks. Traditional jobs disappearing isn't the challenge we face, he argued. Instead it's about upskilling people in all jobs in the tasks that machines can't do (creative work, personal interactions), while also ensuring we have enough expertise to simultaneously build the machines to support the AI that will be entering every industry. How? By changing what we teach (not just the stuff that machines can do just as well), how we teach it (we’re not using the tech we've got well enough in the education sector) and when we teach it (we need lifelong learning, not a front-ended system that spits us out of education never to return).
Cassie Kozyrkov
Melissa Terras took to the stage and declared she was there to persuade us that arts, humanities and social scientists have interesting things to contribute to data science. I was already convinced but keen to hear the case again! She spoke of being discarded for not being tech enough', ignored for not being scientific enough, and then flew the flag for the humanities without stopping for breath. "If you're working with data then you need a sociologist on your team!" Terras' take is that academics from these disciplines are "really good at the ethics stuff" and need to be at the table when data science is happening. She highlighted the UCL Transcribing Bentham project, which among other things crowdsourced more than 60 volunteer citizens to read and translate 8 million handwritten words. The result is full digitization of a massive archive, a now searchable resource of written data and the beginnings of an AI program that can accurately read historical handwritten text.
“If you’re working with tech you need a sociologist on your team!”
Google's Chief Decision Scientist Cassie Kozyrkov referred to herself as 'Captain Obvious', here to tell us a load of things that are really obvious in hindsight. A cerebral and at times surreal session, Kozyrkov tackled some of the most common questions she's asked such as “Is data science a bubble?” (clue: no, it's not) and “What is data science?” (Her answer: the science of making data useful). Part data science 101, part stand-up comedy, she waded right in to some niche but amusing arguments between data engineers, data analysts and statisticians and concluded that what we need above all else is skilled leaders in data science. "People should be able to say they want to be a data scientist and do something else as well…. the other skills matter just as much."
Data Summit was riveting, and I was hugely impressed with how women dominated the agenda from start to finish. Having said that, there is a need for more diversity and intersectionality among the women represented. It would also be encouraging to see more academics among the delegates too, as there is plenty here for us to learn from and take back to the classroom. I'll definitely be back next year to see how we’ve progressed and what new challenges are emerging.
Check out our Twitter Moment for more highlights from DataFest19