Diversity in data science and the legacy of Alan Turing

Hi everyone! Here are my slides from my Lunch and Learn event today, hosted by the Macquarie Group Pride network. They represent my opinions and not (necessarily) those of the Turing or Macquarie Group (although I hope there’s at least some overlap here!)

I was asked to speak about the Life and legacy of Alan Turing and my talk has been heavily inspired by a fantastic panel discussion that I took part in last summer. Hosted by the British Library as part of their GAY UK exhibition, I shared a stage with Prof Andrew Hodges (the man who literally wrote the book on Turing’s life) and Sir John Dermot Turing (Alan Turing’s nephew) - discussed his life and particularly put it in the context of the laws criminalising homosexuality of the time.

Although I’ll include some facts about Turing’s life in my talk today, I want to acknowledge that I’m mostly parroting Andrew and Dermot’s expertise from that fantastic conversation, along with everything I learned at my excellent birthday trip to Bletchley Park last year!

My role on the British Library panel was to represent the Turing Institute and current working data scientists, and my talk today does exactly that. Importantly I’ll cover some key reasons that although homosexuality is no longer illegal in the UK, we still need better representation in data science and artificial intelligence.

I’m the chair of the Turing Ethics Advisory Group and one of the key areas of focus for many of us at the Turing is making sure that technology is built by representative communities. To quote my friend Teon Brooks (co-founder of the fantastic Queer Hack):

“You don’t need a diversity committee, if you have a diverse committee.”

My talk covers some examples of major data science fails:

That last one is particularly appropriate for a talk during LGBT history month. It’s wrong in a bunch of different ways: the experimental psychologist in me is horrified by the terrible study design, the training I’ve received in experiments on human subjects screams out that the participants did not give informed consent, and - maybe most importantly - this is an example of a new technology that can very easily be appropriated to harm LGBTQ+ people.

I work with Josh Cowls who wrote this excellent blog post on AI’s “Trolley Problem” Problem. I’m including it in this talk to highlight the question of who is responsible for ensuring the safety of new technology? In the case of self driving cars, is it the company who builds them? The people who write the algorithm? Governments? Activists? You?

The larger point I’m making is that we need to be building technology, or conducting research that is by the people for the people. I hope that Alan Turing inspires many LGBTQ+ people to achieve their goals in science, technology, engineering, maths and medicine (STEMM) and beyond. But as we progress it is imperative that we look around and consider whose voices are left out of our conversations? Are we supporting people who lie at the intersections of traditionally underrepresented groups? How can we build inclusive communities, together?

A very non-comprehensive list of references for people wanting to know more and spread the word:

My slides are licensed CC-BY which means you can use them (and remix them) for anything you’d like so long as you credit me. You can do that by citing this DOI: 10.6084/m9.figshare.5932579. I’d love to know if you find them helpful. 🌈💖🙌