ABSTRACT
In this paper we argue that standard calls for explainability that focus on the epistemic inscrutability of black-box machine learning models may be misplaced. If we presume, for the sake of this paper, that machine learning can be a source of knowledge, then it makes sense to wonder what kind of \em justification it involves. How do we rationalize on the one hand the seeming justificatory black box with the observed wide adoption of machine learning? We argue that, in general, people implicitly adoptreliabilism regarding machine learning. Reliabilism is an epistemological theory of epistemic justification according to which a belief is warranted if it has been produced by a reliable process or method \citegoldman2012reliabilism. We argue that, in cases where model deployments require \em moral justification, reliabilism is not sufficient, and instead justifying deployment requires establishing robust human processes as a moral "wrapper'' around machine outputs. We then suggest that, in certain high-stakes domains with moral consequences, reliabilism does not provide another kind of necessary justification---moral justification. Finally, we offer cautions relevant to the (implicit or explicit) adoption of the reliabilist interpretation of machine learning.
- Kai Arulkumaran, Antoine Cully, and Julian Togelius. 2019. Alphastar: An evolutionary computation perspective. arXiv preprint arXiv:1902.01724 (2019).Google Scholar
- Muzammil Bashir. 2019. Deep Learning Approach to Trespass Detection using Video Surveillance Data. (2019).Google Scholar
- Ruha Benjamin. 2019. Race after technology: Abolitionist tools for the new jim code. John Wiley & Sons.Google Scholar
- Richard Berk. 2019. Machine learning risk assessments in criminal justice settings. Springer.Google Scholar
- Reuben Binns, Max Van Kleek, Michael Veale, Ulrik Lyngs, Jun Zhao, and Nigel Shadbolt. 2018. 'It's Reducing a Human Being to a Percentage': Perceptions of Justice in Algorithmic Decisions. In Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems. ACM, 377.Google ScholarDigital Library
- Joanna J Bryson, Mihailis E Diamantis, and Thomas D Grant. 2017. Of, for, and by the people: the legal lacuna of synthetic persons. Artificial Intelligence and Law 25, 3 (2017), 273--291.Google ScholarDigital Library
- Jenna Burrell. 2016. Howthe machine "thinks": Understanding opacity in machine learning algorithms. Big Data & Society 3, 1 (2016), 2053951715622512.Google ScholarCross Ref
- Richmond Campbell. 2015. Moral Epistemology. In The Stanford Encyclopedia of Philosophy (winter 2015 ed.), Edward N. Zalta (Ed.). Metaphysics Research Lab, Stanford University.Google Scholar
- Alexandra Chouldechova. 2017. Fair prediction with disparate impact: A study of bias in recidivism prediction instruments. Big data 5, 2 (2017), 153--163.Google Scholar
- Mark Coeckelbergh. 2019. Artificial Intelligence, Responsibility Attribution, and a Relational Justification of Explainability. Science and engineering ethics (2019), 1--18.Google Scholar
- Earl Conee and Richard Feldman. 1998. The generality problem for reliabilism. Philosophical Studies 89, 1 (1998), 1--29.Google ScholarCross Ref
- Finale Doshi-Velez and Been Kim. 2017. Towards a rigorous science of interpretable machine learning. arXiv preprint arXiv:1702.08608 (2017).Google Scholar
- Laurel Eckhouse, Kristian Lum, Cynthia Conti-Cook, and Julie Ciccolini. 2019. Layers of bias: A unified approach for understanding problems with risk assessment. Criminal Justice and Behavior 46, 2 (2019), 185--209.Google ScholarCross Ref
- Andre Esteva, Alexandre Robicquet, Bharath Ramsundar, Volodymyr Kuleshov, Mark DePristo, Katherine Chou, Claire Cui, Greg Corrado, Sebastian Thrun, and Jeff Dean. 2019. A guide to deep learning in healthcare. Nature medicine 25, 1 (2019), 24--29.Google Scholar
- Richard Feldman. 2003. Epistemology. (2003).Google Scholar
- Edmund Gettier. 1963. Is justified true belief knowledge? 1963 (1963), 273--274.Google Scholar
- Alvin Goldman and Bob Beddor. 2016. Reliabilist Epistemology. In The Stanford Encyclopedia of Philosophy (winter 2016 ed.), Edward N. Zalta (Ed.). Metaphysics Research Lab, Stanford University.Google Scholar
- Alvin I Goldman. 2012. Reliabilism and contemporary epistemology: essays. Oxford University Press.Google Scholar
- David Gunning. 2017. Explainable artificial intelligence (xai). Defense Advanced Research Projects Agency (DARPA), nd Web 2 (2017).Google Scholar
- Lars Hall, Petter Johansson, Betty Tärning, Sverker Sikström, and Thérèse Deutgen. 2010. Magic at the marketplace: Choice blindness for the taste of jam and the smell of tea. Cognition 117, 1 (2010), 54--61.Google ScholarCross Ref
- Andreas Holzinger, Chris Biemann, Constantinos S Pattichis, and Douglas B Kell. 2017. What do we need to build explainable AI systems for the medical domain? arXiv preprint arXiv:1712.09923 (2017).Google Scholar
- Gábor Horváth, Balázs Bernáth, and Gergely Molnár. 1998. Dragonflies find crude oil visually more attractive than water: multiple-choice experiments on dragonfly polarotaxis. Naturwissenschaften 85, 6 (1998), 292--297.Google ScholarCross Ref
- Jochen Kruppa, Alexandra Schwarz, Gerhard Arminger, and Andreas Ziegler. 2013. Consumer credit risk: Individual probability estimates using machine learning. Expert Systems with Applications 40, 13 (2013), 5125--5131.Google ScholarCross Ref
- Jarrett Leplin. 2007. In defense of reliabilism. Philosophical Studies 134, 1 (2007), 31--42.Google ScholarCross Ref
- David Lewis. 1996. Elusive knowledge. Australasian journal of Philosophy 74, 4 (1996), 549--567.Google ScholarCross Ref
- Zachary C Lipton, Yu-Xiang Wang, and Alex Smola. 2018. Detecting and correcting for label shift with black box predictors. arXiv preprint arXiv:1802.03916 (2018).Google Scholar
- Lydia T Liu, Sarah Dean, Esther Rolf, Max Simchowitz, and Moritz Hardt. 2018. Delayed impact of fair machine learning. arXiv preprint arXiv:1803.04383 (2018).Google Scholar
- Spyros Makridakis. 2017. The forthcoming Artificial Intelligence (AI) revolution: Its impact on society and firms. Futures 90 (2017), 46--60.Google ScholarCross Ref
- Gary Marcus. 2018. Deep learning: A critical appraisal. arXiv preprint arXiv:1801.00631 (2018).Google Scholar
- Deirdre K Mulligan, Daniel Kluttz, and Nitin Kohli. 2018. Contestability and Professionals: From Explanations to Engagement with Algorithmic Systems. Available at SSRN 3311894 (2018).Google Scholar
- Harold Noonan. 2002. Routledge philosophy guidebook to Hume on knowledge. Routledge.Google Scholar
- William J Prior. 2016. Virtue and knowledge: An Introduction to ancient Greek ethics. Routledge.Google Scholar
- Stathis Psillos. 2007. Philosophy of science AZ. Edinburgh University Press.Google Scholar
- Stathis Psillos. 2014. Causation and explanation. Routledge.Google Scholar
- Willard Van Orman Quine and Joseph Silbert Ullian. 1978. The web of belief. Vol. 2. Random House New York.Google Scholar
- Conrad Sachweh. 2018. General Data Protection Regulation and Explainable Machine Learning Challenges. (2018).Google Scholar
- Wojciech Samek, Thomas Wiegand, and Klaus-Robert Müller. 2017. Explainable artificial intelligence: Understanding, visualizing and interpreting deep learning models. arXiv preprint arXiv:1708.08296 (2017).Google Scholar
- Andrew D Selbst and Solon Barocas. 2018. The intuitive appeal of explainable machines. Fordham L. Rev. 87 (2018), 1085.Google Scholar
- Andrew D Selbst, Danah Boyd, Sorelle A Friedler, Suresh Venkatasubramanian, and Janet Vertesi. 2019. Fairness and abstraction in sociotechnical systems. In Proceedings of the Conference on Fairness, Accountability, and Transparency. ACM, 59--68.Google ScholarDigital Library
- Stefan Thaler, Vlado Menkovski, and Milan Petkovic. 2018. Deep Learning in Information Security. arXiv preprint arXiv:1809.04332 (2018).Google Scholar
- Jeremy Waldron. 2009. Judges as moral reasoners. International Journal of Constitutional Law 7, 1 (2009), 2--24.Google ScholarCross Ref
- Douglas Walton. 2010. Appeal to expert opinion: Arguments from authority. Penn State Press.Google Scholar
Index Terms
- Why Reliabilism Is not Enough: Epistemic and Moral Justification in Machine Learning
Recommendations
Percentages and reasons: AI explainability and ultimate human responsibility within the medical field
AbstractWith regard to current debates on the ethical implementation of AI, especially two demands are linked: the call for explainability and for ultimate human responsibility. In the medical field, both are condensed into the role of one person: It is ...
Organization Studies and the New Pragmatism: Positivism, Anti-Positivism, and the Search for Ethics
The central claim of this paper is that organization studies needs to be fundamentally reshaped. Such change is needed to provideroom for ethics and to increase the relevance of research. We argue that the new pragmatism provides critical resources ...
The relationship between the uniqueness ofcomputer ethics and its independence as adiscipline in applied ethics
A number of different uniqueness claims have been made about computer ethics in order to justify characterizing it as a distinct subdiscipline of applied ethics. I consider several different interpretations of these claims and argue, first, that none are ...
Comments