Roundup: panel meeting on data ethics
Hugh Lawson-Tancred, Ethics Co-ordinator at the Society of Data Miners and Ethics for Analytics Group Leader, shares the highlights from the latest London Business Analytics meet up discussing data ethics and data mining.
Last month the London Business Analytics meet up group held a panel discussion to discuss a wide-range of topics around the ethics of data mining and machine learning.The panel consisted of five industry experts: Alexandru Agachi, COO of Empiric Capital; Andrew Fryer, Big Data Expert at Microsoft; Chiara Grattini, Cultural Anthropologist at Intel; Steph Locke, Principal Consultant at Mango Solutions; and myself. Along with the five panellists, there were 150 guests attending.
The panel covered a large number of issues around consent, disclosure and reuse. One possibility raised was that of modelling the development of consent to data processing and the evolution of clinical consent. This model would enable the benefits of a mature consent practice to be carried over. There was also a discussion around restricting the reusability of data, especially in the crucial area of genetic data. This was connected to the issue of the provision of data to the NHS and its right to use such data in the public interest.
The audience raised a question concerning the effect of increased data use for medical insurance and the societal treatment of individuals with expensive health requirements. This prompted a consideration of the advantages of selective opting in rather than ‘all in’ or an ‘all out’ of a service, irrespective of the administrative complexity.
Attitudes to privacy
The discussion also looked at the role of personal data in retail finance. Attitudes to privacy are most conservative among older age groups, raising the question as to whether and to what extent people are prepared to accept an intrusion into their privacy in return for a service. The question of the complexity and sheer length of consent agreements was raised, but one panellist was adamant that there had been significant improvements in this area over the last few years.
Two panellists raised the key issue of de-identification and the threat it posed to the prevailing assumption of the safety of aggregated and anonymised data. There was also a vigorous discussion about what data professionals need to be mindful of when working on data that could have a privacy angle.
Although the open session ended without a clear consensus being reached, the general feeling in the post-meeting discussion was that this relatively unusual format had been a notable success. A number of important issues and lines of thought had been broached in what is an increasingly controversial, but still in many ways, nebulous problem area for the digital age.
Hugh Lawson-Tancred is Ethics Co-ordinator at the Society of Data Miners and Ethics for Analytics Group Leader of the Personal Data & Trust Network.
Check out the latest Personal Data and Trust Network news here.