Mapping the Stony Road toward Trustworthy AI: Expectations, Problems, Conundrums

In: Marcello Pelillo and Teresa Scantamburlo (Eds.). Machines We Trust: Perspectives on Dependable AI. Cambridge, MA: MIT Press, Forthcoming

14 Pages Posted: 29 Dec 2020

See all articles by Gernot Rieder

Gernot Rieder

University of Hamburg - Department of Informatics

Judith Simon

University Hamburg -Department of Informatics; University of Hamburg - Department of Philosophy

Pak-Hang Wong

University of Hamburg - Department of Informatics

Date Written: October 23, 2020

Abstract

The notion of trustworthy AI has been proposed in response to mounting public criticism of AI systems, in particular with regard to the proliferation of such systems into ever more sensitive areas of human life without proper checks and balances. In Europe, the High-Level Expert Group on Artificial Intelligence has recently presented its Ethics Guidelines for Trustworthy AI. To some, the guidelines are an important step for the governance of AI. To others, the guidelines distract effort from genuine AI regulation. In this chapter, we engage in a critical discussion of the concept of trustworthy AI by probing the concept both on theoretical and practical grounds, assessing its substance and the feasibility of its intent. We offer a concise overview of the guidelines and their vision for trustworthy AI and examine the conceptual underpinnings of trustworthy AI by considering how notions of 'trust' and 'trustworthiness' have been discussed in the philosophical literature. We then discuss several epistemic obstacles and moral requirements when striving to achieve trustworthy AI in practice before concluding with an argument in support of the establishment of a trustworthy AI culture that respects and protects foundational values.

Keywords: Trust, Trustworthy AI, Fairness, Transparency, Accountability, AI Ethics

Suggested Citation

Rieder, Gernot and Simon, Judith and Wong, Pak-Hang, Mapping the Stony Road toward Trustworthy AI: Expectations, Problems, Conundrums (October 23, 2020). In: Marcello Pelillo and Teresa Scantamburlo (Eds.). Machines We Trust: Perspectives on Dependable AI. Cambridge, MA: MIT Press, Forthcoming, Available at SSRN: https://ssrn.com/abstract=3717451 or http://dx.doi.org/10.2139/ssrn.3717451

Gernot Rieder (Contact Author)

University of Hamburg - Department of Informatics ( email )

Vogt-Koelln-Strasse 30
Hamburg, 22527
Germany

Judith Simon

University Hamburg -Department of Informatics ( email )

Vogt-Koelln-Strasse 30
Hamburg, 22527
Germany

University of Hamburg - Department of Philosophy ( email )

Von-Melle-Park 6
Hamburg, D-20146
Germany

Pak-Hang Wong

University of Hamburg - Department of Informatics ( email )

Vogt-Koelln-Strasse 30
Hamburg, 22527
Germany

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
144
Abstract Views
779
rank
257,068
PlumX Metrics