Call for Evidence
Algorithms are being used to make decisions in a growing range of contexts, from mortgages to sentencing criminals. How an algorithm is formulated, its scope for error or correction, the impact it may have on an individual – and their ability to understand or challenge that decision - are increasingly relevant questions. A call for evidence was made on 28 February 2017, and followed the Committee’s work on Robotics and AI, and its call for a standing Commission on Artificial Intelligence.
Summary of the Turing’s submission
The submission proposes the need for further investment for research into the technical, ethical, and legal challenges surrounding algorithms in decision-making, in which consideration of the interplay between automated and human decision-making will be crucial.
Also proposed is the scoping of an oversight institution for algorithms, e.g. an independent regulatory body to monitor and hold to account organisations which use algorithms to make decisions. It would be of the utmost importance that such a body, or bodies, that would ‘decide’ what is fair, unbiased, and transparent in the context of algorithmic decision-making, are themselves working in the best interests of the many and not the few.
Recommendation is made for the Committee to consider a ‘right to explanation’ of automated decisions, the role of human interpretation when decisions involve sensitive or personal information, and the establishment of certification mechanisms that will test algorithms for possible unethical consequences prior to their deployment.
Luciano Floridi, Chair of the Alan Turing Institute’s Data Ethics Group, Turing Fellow, University of Oxford
Jonathan Cave, Department of Economics, University of Warwick
Jennifer Davis, Faculty of Law, University of Cambridge
Brent Mittelstadt, Turing Research Fellow, UCL
Charles Raab, Turing Fellow, University of Edinburgh
Sandra Wachter, Turing Fellow, Oxford Internet Institute, University of Oxford
Adrian Weller, Turing Fellow, University of Cambridge
Robert Maskell, Director of High Performance Computing at Intel