Event report by Giles Pavey, RSS Data Science Section Committee member
Wednesday March 17th saw the RSS Data Science Section host its first ‘Ethics Happy Hour’. Events in this new series provide an opportunity to discuss and meet other people interested in questions of AI ethics and data science ethics more broadly. Taking place in a relaxed and informal setting, our aim for these sessions is to stimulate intellectual exchange and contribute to community building around ethics in the context of data science.
The inaugural event took place virtually and focused on COVID-19. The discussion took the form of a panel chaired by RSS Data Science Section Committee member Dr Florian Ostmann with three experts sharing their thoughts on the ethics of data science in addressing the public health crisis:
- Dr Zachary Lipton (Carnegie Mellon University) is a machine learning researcher and jazz saxophonist. He is currently an Assistant Professor of Operations Research and Machine Learning at Carnegie Mellon University, where he runs the Approximately Correct Machine Intelligence lab.
- Dr Anjali Mazumder (RSS Data Science Section Committee / The Alan Turing Institute) is the Theme Lead on AI and Justice & Human Rights at the Alan Turing Institute and also a member of the Data Science Section Committee, among other RSS roles. Her research interests include Bayesian decision support systems, causal reasoning, detecting bias and algorithmic fairness, and responsible data sharing practices.
- Dr Nicola Stingelin (RSS Data Ethics and Governance Section Committee) is a member of the RSS Data Ethics and Governance Section Committee and an associate researcher at the University of Basel. Building on business experience in the pharmaceutical sector, she acts in various advisory roles with a focus on the ethics of data innovation including big data, algorithms and public health data ethics in health care research and practice.
The event attracted around 35 attendees from across academia, business, and the public and charitable sectors. After the introduction there was a lively debate covering multiple aspects of the use of data and AI around the world in response to the pandemic and its ramifications.
Access to data was a major discussion point, with Nicola arguing that the importance of data in creating competitive advantage, especially in commerce, was causing obstacles to data sharing which could stall progress.
Zach conjectured that we should not rely on technical solutions to what might be considered primarily societal and philosophical problems such as access to data or AI resource. In answer to this point, Anjali suggested that there are at least some relevant problems where technical solutions can help. For instance, Privacy Enhancing Technologies, such as Differential Privacy, can enable insights whilst protecting individuals’ privacy.
Another area of debate concerned differences between what we should expect from data and AI when thinking about micro (individual) level predications and classifications versus more high/macro-level decision making. For example, is it acceptable to use thermal imaging AI models in public spaces for social control?
The discussion also touched on areas such as whether society is becoming too reliant on data and whether this is creating a digital divide; how we feel about a society where one has to have a smart phone to access services; and how things will develop as the populous are asked to share more and more data. Lastly, on the subject of speed of change: COVID19 has highlighted tensions between the desire and need to move at a quick pace and the academic norm of considered peer review.
The event was drawn to a close (having overrun by 15’) with a general agreement that the ethical issues are many and complex and that data science and statistical methods will offer key tools for us to navigate the future.
Please contact the RSS Data Science Section if you have any comments or would like to suggest topics for future events via email to firstname.lastname@example.org. To stay informed about future ethics happy hours and other events organised by the Data Science Section, we recommend signing up to the Data Science Section mailing list.