An ethics book club in a box
This article was written by Christine Henry, with additional contributions by Ruby Childs, DataKind UK Ethics Committee
In 2019, DataKind UK set up a data science ethics book club, with support from Open Society Foundations. Over the course of the year, we ran seven book club sessions, with a total attendance of over 200. They ran every 6–8 weeks, beginning in London and growing to include sessions in Edinburgh and online.
We’re opening up our resources on materials and processes, so that others interested in discussing the ethics of data and AI in society can set up their own groups too!
Why a book club?
We set out three aims for the book club:
- To foster discussions of ethical issues among our data science community and beyond, that they can take back to their day jobs/careers;
- To educate and inform our community on ethical issues that may come up in data science projects, to better enable them to support social change organisations;
- To strengthen our community of data scientists, social change organisations, and data for good enthusiasts.
How do we broaden access to the discussions?
We understand not everyone can make time to read a lengthy book, and the cost of books can also be a barrier for some people. So in order to ensure we cater for a range of people and increase accessibility, we went beyond just recommending a book and instead provided reading lists with content of varying formats, durations and technical content. This included journal articles, blogs, news stories, videos and poems.
This may seem unconventional for a ‘book’ club, but it increased accessibility while also encouraging richer, more diverse conversations as participants had a wide range of sources to reflect upon. Retrospectively, it has been successful in exposing people to different ideas and in increasing participation — we recommend looking beyond non-fiction books and peer-reviewed articles for others looking to discuss data ethics as well. (For those interested in a book club covering the same ground using fiction, Casey Fiesler and team have you covered!)
One of our goals is to increase awareness of ethics and AI within wider society including our civil society partners. The book club is not only for our existing community of data scientists but for anyone interested in data ethics, including those who are not in technical roles. To encourage participants from a variety of backgrounds, we had a mixture of technical and non-technical reading material and aimed to include “primer”-type materials on tech where possible, as well as more impact-focused pieces.
As the book club became more popular, attracting 40+ people to a single session, we made sure the discussion groups stayed under 10 people (ideally 6–8), to ensure that everyone could meaningfully participate if they desired. We sought and trained facilitators to lead each of these discussion groups.
Process please!
We broke our book club down into a number of roles: hosting, facilitating and participating. Our hosts really led and begun the book club, whilst our facilitators encouraged diverse, inclusive and thoughtful conversation. Through these roles, we were able to develop recommended processes.
- Book Club FAQs: The FAQs were included in each book club event invitation, and provides participants a deeper understanding of the book club. Telling people up front about objectives, audience, costs, and any requirements for attendance can help prospective newcomers feel welcome.
- Facilitator Guide: This guide provides direction for the facilitators, from welcoming participants, starting conversations and how to ensure inclusive discussions. In particular, we found it particularly helpful to provide facilitators with a set of suggested questions and conversation starters specific to each topic.
- Host Guide: This guide is particularly useful especially when moving hosts and venues regularly. It’s important to make sure all participants have a shared understanding of the house rules — such as the Chatham House rules which we regularly follow.
Our topics for 2019
The DataKind UK Ethics committee chose topics and reading materials. These were their selections for 2019. They may help you get started too!
- Impacts of Algorithms
- Facial Recognition
- Fairness in AI
- Self-Driving Cars
- AI and Gender
- AI and Financial Inclusion
- AI and Race
All the materials used in the DataKind UK Data Ethics Book club can be found on our GitHub repo.
Suggestions, support, questions
As with all our work, we aim to be as open as possible. Our ‘book club in a box’ is available on GitHub. Please do replicate — and, we hope, improve upon! — our approach. Get in touch at contact@datakind.org.uk if you’ve got any questions.
Join in!
The book club continues in 2020! The session of the year is on recommender systems, taking place in London, Edinburgh and online:
Keep an eye out for future local and remote events here or sign up to get email updates from DataKind UK.
