"Meaningful Information" and the Right to Explanation

By Andrew D. Selbst, Julia E. Powles. 2017

451 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Law, Sociology and Political Science

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Law

In Sociology and Political Science

Citations in the corpus

Data Governance in the Age of Large-Scale Data-Driven Language Technology (2022, background): “14GDPR [49] does not explicitly guarantee it, but it it does require the data processor to provide ‘meaningful information about the logic involved’ in fully automated decisions, which could be interpreted that way [126].”
Algorithms that remember: model inversion attacks and data protection law (2018, background): “The nature and utility of these provisions is explored and debated extensively elsewhere [53, 20, 44, 29]; for reasons of space we do not expand upon them here.”
The Bouncer Problem: Challenges to Remote Explainability (2019, background): “This translated in some jurisdictions ( e.g. , United States of America and Europe) into a so called right to explanation [12,26], that states that the output decisions of an algorithm must be motivated.”
TED: Teaching AI to Explain its Decisions (2018, result): “…result has been observed by other researchers who have suggested adding “ra-tionales” to improve classiﬁer performance, but not for ex-plainability (Sun and DeJong 2005; Zaidan and Eisner 2007; Zaidan and Eisner 2008; Zhang, Marshall, and Wallace 2016; McDonnell et al. 2016; Donahue and Grauman…”
Your evidence? Machine learning algorithms for medical diagnosis and prediction (2019, background): “Some legal scholars (e.g., Selbst & Powles, 2017) even argue that the GDPR includes a “ right to an explanation ” while others (e.g., Wachter, Mittelstedt, & Floridi, 2017) reject such a reading as too strong.”
The hidden assumptions behind counterfactual explanations and principal reasons (2019, unknown): “sions, and, if it exists, whether and when it requires an explanation of specific decisions or the model [4, 5, 10, 19, 26, 28, 39, 46].”
One Explanation Does Not Fit All: A Toolkit and Taxonomy of AI Explainability Techniques (2019, background): “This has led to some ambiguity in regulations [24, 45, 50, 56] and researchers defining their own problem statements, as well as solutions.”
Legal requirements on explainability in machine learning (2020, background): “The majority of scholars support this requirement of explanation (Goodman and Flaxman 2016; Malgieri and Comandé 2017; Edwards and Veale 2018; Selbst and Powles 2017).”
Explainable AI under contract and tort law: legal incentives and technical challenges (2020, background): “…countries, because of the strong varieties of capitalism element in corporate mergers, for instance Japan, partly still before machine learning, partly with machine learning based on a large-scale data set concentrating on the Japanese corporate setting (Shibayama 2008; Shao et al. 2018).”
Translation Tutorial : A Shared Lexicon for Research and Practice in Human-Centered Software Systems (2018, background): “Policymakers are elevating explanations to the status of individual rights in several jurisdictions (Selbst and Powles, 2017), but often without defining what would constitute an explanation or how it might be achieved.”
Interpretable Machine Learning (2019, background): “However, some academics have raised concerns that this is not enforceable in practice, as the recitals are not legally binding.(32,77,81) The ICO has stated that an individual must be given an explanation to enable their right to receive meaningful information about a decision.”
Algorithms to estimate Shapley value feature attributions (2022, background): “This demand is reﬂected in increasing calls for explanations by diverse regulatory bodies, such as the General Data Protection Regulation’s “right to explanation” [12] and the Equal Credit Opportunity Act’s adverse action notices [13].”
The Fallacy of AI Functionality (2022, background): “There has been much discussion about whether the GDPR includes a "right to explanation" and what such a right entails [50, 103, 167, 193].”
The fallacy of inscrutability (2018, background): “Most often, this comes as a demand that systems be able to produce explanations of their behaviours [2,12,13].”