This Royal Statistical Society has published a report summarising the discussions that took place during a two-day consultation on the opportunities and ethics of big data.
The consultation workshop, which took place in November 2015 at St George’s House in Windsor, brought together key figures in government, academia and the private sector, to explore the issues.
The event was chaired by Denise Lievesley, former RSS president and principal at Green Templeton College in Oxford. Among the 28 participants were Bernard Silverman, the Home Office chief scientific advisor, Howard Covington, chair of the new Alan Turing Institute and Paul Maltby, director for data at the Government Digital Service.
The event kicked-off by identifying the opportunities presented by big data – in particular that it provides a new context for statistical analysis, boosting its appeal and application to wide-ranging problems. The event highlighted big data’s potential to change the type of evidence used by policy makers, as data visualisation, computer models and predictive analytics can now be used as a basis for decision making.
The workshop then went on to address the challenges created by these new opportunities, including whether current ethical frameworks are fit for purpose. Key concerns were highlighted such as trust in big data methods and algorithms, and avoiding possible risks to individuals.
The workshop’s participants agreed a set of guiding principles regarding the use of big data in general. The principles focused around transparency, fairness, robustness of data (particularly when being adapted for other uses) and ensuring that the benefits of using big data are fully explored and recognised in decision-making processes.
Participants also agreed that improvements are needed in skills, analytical capabilities, training and public engagement. Governance and leadership could be addressed by establishing an independent Council for Big Data Ethics to help formulate and uphold an authoritative ethical framework.
The ethical issues raised by big data are highly dependent on context, so data controllers/owners might need to use ad hoc policies and ‘influence in practice’ to address concerns. The workshop participants also recognised that wider public consultation is needed to further explore public views on big data.
Further discussion of the workshop’s findings, as well as a full list of the participants can be found in the full report (available as a PDF).