About | HeinOnline Law Journal Library | HeinOnline Law Journal Library | HeinOnline
LOG IN
LOG IN

31 Harv. J. L. & Tech. 841 (2017-2018)
Counterfactual Explanations without Opening the Black Box: Automated Decisions and the GDPR

handle is hein.journals/hjlt31 and id is 859 raw text is: Harvard Journal of Law & Technology Volume 31, Number 2 Spring 2018 COUNTERFACTUAL EXPLANATIONS WITHOUT OPENING THE BLACK Box: AUTOMATED DECISIONS AND THE GDPR Sandra Wachter, * Brent Mittelstadt, * * & Chris Russell* ** TABLE OF CONTENTS I. IN TR O D U CTIO N .............................................................................. 842 II. C OUN TERFACTUALS ..................................................................... 844 A. Historic Context and the Problem of Knowledge ..................... 846 B. Explanations in A.I andMachine Learning ............................ 849 C. Adversarial Perturbations and Counterfactual E xp lan a tion s .......................................................................... 85 1 D . Causality and F airness ............................................................ 853 III. GENERATING COUNTERFACTUALS ............................................. 854 A . L SA T D ataset ........................................................................... 856 B. Pim a D iabetes D atabase .......................................................... 859 C. Causal Assumptions and Counterfactual Explanations ........... 859 IV. ADVANTAGES OF COUNTERFACTUAL EXPLANATIONS ............... 860 V. COUNTERFACTUAL EXPLANATIONS AND THE GDPR .................. 861 * Oxford Internet Institute, University of Oxford, 1 St. Giles, Oxford, OXI 3JS, UK and The Alan Turing Institute, British Library, 96 Euston Road, London, NW1 2DB, UK. E- mail: sandra.wachterttoii.ox.ac.uk. This work was supported by The Alan Turing Institute under the EPSRC grant EP/N510129/1. ** Oxford Internet Institute, University of Oxford, 1 St. Giles, Oxford, OXI 3JS, UK, The Alan Turing Institute, British Library, 96 Euston Road, London, NW1 2DB, UK, Department of Science and Technology Studies, University College London, 22 Gordon Square, London, WC1E 6BT, UK. *** The Alan Turing Institute, British Library, 96 Euston Road, London, NW1 2DB, UK, Department of Electrical and Electronic Engineering, University of Surrey, Guildford, GU2 7HX, UK. The authors would like to thank Patrick Allo, Dan Burk, Lee Bygrave, Mireille Hilde- brandt, David Landsberg, Zachary Lipton, Alessandro Spina, David Sutcliffe, Jade Thomp- son, Jessie Wand, Tal Zarsky, participants of the 2018 Privacy Law Scholars Conference (Brussels), the 2017 Ethical Auditing in Automated Decision-Making workshop (Oxford), and the 2018 Artificial Intelligence, Ethics and the Law workshop (London), and members of the Alan Turing Institute's Fairness, Transparency and Privacy interest group for their invaluable feedback and comments. We would also like to thank the Oxford Internet Insti- tute and Alan Turing Institute for providing the highly interdisciplinary environments in which the authors were able to develop the central idea of the paper. Last but not least we want to thank the editorial team of Harvard Journal of Law and Technology for their invalu- able feedback and insightful comments that massively improved the quality of our work.

What Is HeinOnline?

HeinOnline is a subscription-based resource containing thousands of academic and legal journals from inception; complete coverage of government documents such as U.S. Statutes at Large, U.S. Code, Federal Register, Code of Federal Regulations, U.S. Reports, and much more. Documents are image-based, fully searchable PDFs with the authority of print combined with the accessibility of a user-friendly and powerful database. For more information, request a quote or trial for your organization below.



Short-term subscription options include 24 hours, 48 hours, or 1 week to HeinOnline.

Purchase Access

Contact us for annual subscription options:

Contact Us

Already a HeinOnline Subscriber?

Log In

profiles profiles most