CoChat

Slave to the Algorithm? Why a Right to Explanationn is Probably Not the Remedy You are Looking for

Open access
Edwards, Lilian, Veale, Michael · 2017 · SSRN Electronic Journal

The DOI matches its Crossref record; 11 of the 11 references it cites were found in a registry or an index.

Overview

Integrity signals

Registry

Crossref · 10.2139/ssrn.2972855

Cites

11 of 11 references verified0 citation errors · 0 not checkable

Statistics

Recomputing p-values needs the paper’s full text.

Verification

Claims the paper makes

We couldn’t retrieve this paper’s full text, so its claims and self-check weren’t run. The registry and citation checks still apply.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Citations

Cites

Show all 11 references

Research works printed but not linked: 29

  1. Khazan, 2015, How to Fake Your Workout, THE ATLANTIC
  2. 2014, The Scored Society: Due Process for Automated Predictions, 89 WASH, L. REV, 1
  3. This has been interpreted to imply that European systems are more interested in the human dignity of data subjects than the US system
  4. ; See Henriette & Niva Elkin-Koren, 2015, The Effects of Transparency on Trust in and Acceptance of a Content-Based Art Recommender, 18 USER MODELING AND USER-ADAPTED INTERACTION 455 (2008); while others consider trust building overall. See contra A Busone et al., The Role of Explanations on Trust and Reliance in Clinical Decision Support Systems, ICHI '15, 160
  5. Matthew Chalmers, 2003, Seamful and Seamless Design in Ubiquitous Computing, WORKSHOP AT THE CROSSROADS: THE INTERACTION OF HCI AND SYSTEMS ISSUES IN UBICOMP
  6. Bruno See, 1999, ESSAYS ON THE REALITY OF SCIENCE STUDIES, 304
  7. Motahhare Eslami, 2016, First I "Like" It, Then I Hide It: Folk Theories of Social Feeds
  8. 2008, that ML algorithms such as search algorithms are protected as trade secrets. See especially Viacom Intern, 259
  9. 2011, For counter methods, see Ananad D. Sarwate et al., Signal Processing and ML with Differential Privacy: Algorithms and Challenges for Continuous Data, See the literature on model inversion attacks, including Matt Fredrikson et al. Model Inversion Attacks that Exploit Confidence Information and Basic Countermeasures, 86
  10. Ji Zhanglong, 2014, Differential Privacy and ML: A Survey and Review, 10.7554/elife.02451.006 · doi:10.7554/elife.02451.006
  11. Larry See, 2002, Basic Books, 2006); see also visions of this in the marketing literature, such as ALAN MITCHELL, RIGHT SIDE UP, 6
  12. Richard Mortier, 2010, The Personal Container or Your Life in Bits, DIGITAL FUTURES '10
  13. Hildebrandt, 10.1007/978-3-663-06806-8_8 · DOI
  14. See infra Section V
  15. This bizarre choice can only be explained by thinking of art. 20 as a solution to promote competition by allowing data subjects to make active choices to retrieve their voluntarily posted data from social networks
  16. Art Gdpr, )(b), 20
  17. Interestingly, she rejects as part of the "opportunity to be heard" a simple right to access to the algorithm's source code and/or a hearing on the logic of its decision as too expensive under the balancing test in Matthews v. Eldridge (Citron
  18. See Tutt, 2016, Other suggestions for algorithmic audit are usefully compiled by Brent Mittelstadt, Auditing for Transparency in Content Personalization Systems, supra note 13, 10
  19. 2014, See The Data Protection Impact Assessment Template supported by Commission Recommendation 2014/724/EU, Smart Grid task Force 2012-14, Expert Group 2: Regulatory Recommendations for Privacy, Data Protection and Cybersecurity in the Smart Grid Environment, Data Protection Impact Assessment Template for Smart Grid and Smart Metering Systems
  20. Art Gdpr, 35, 10.2307/j.ctv6cfnkk.10 · DOI
  21. Ico, 10.1787/888933003744 · doi:10.1787/888933003744
  22. Art Gdpr, 2016, See also DATA PROTECTION WORKING PARTY, Guidelines on Data Protection Impact Assessment (DPIA) and Determining Whether Processing Is "Likely to Result in a High Risk" for the Purposes of Regulation, Art, 35
  23. 2016, In other work, one author has suggested that PIAs could be developed into more holistic Social Impact Assessments (SIAs) and although this was developed to deal with the IoT it might also have considerable application to ML systems: see Lilian Edwards et al., From Privacy Impact Assessment to Social Impact Assessment, IEEE SECURITY AND PRIVACY WORKSHOPS, 53
  24. Rowena See, 2013, Developing a Privacy Seal Scheme
  25. Gemma Farmer, 2015, What's the Latest on the ICO Privacy Seals?
  26. Issues of algorithmic fairness are specifically discussed in GDPR, recital 71. Tristan Henderson in private correspondence has suggested that a certifying authority might well under art
  27. 2015, See Schrems v. Data Prot. Comm'r of Ir
  28. 2000, TrustE and similar privacy seals failed to meet European privacy standards, Data Privacy in Cyber Space, LAW AND THE INTERNET
  29. Art Gdpr, This may lead to a new world of form-filling for data controllers

Cited by

Metadata

Authors
Edwards, L., & Veale, M.
Affiliations
  • University of Newcastle Australia (AU)
  • University College London (GB)
Corresponding author
Edwards, Lilian
Venue
SSRN Electronic Journal
Publisher
Elsevier BV
Published
2017
ISSN
1556-5068 (ISSN-L)
OpenAlex
W2618631885
Cited by
244
References
11
Citations by year
2026: 4 · 2025: 16 · 2024: 23 · 2023: 16 · 2022: 17 · 2021: 55 · 2020: 55 · 2019: 37 · 2018: 20 · 2017: 1
Type
posted content
Language
en
Open access
Locations
SSRN Electronic Journal (publishedVersion)
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › Explainable Artificial Intelligence (XAI)
  • Social Sciences › Law › Digitalization, Law, and Regulation
  • Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
Keywords
right to explanation, algorithmic transparency, General Data Protection Regulation, algorithmic discrimination, algorithmic bias, right to be forgotten, data portability, privacy by design, data protection impact assessment, algorithmic accountability, algorithmic opacity

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.