Fairness in Agreement With European Values

Abstract

With increasing digitalization, Artificial Intelligence (AI) is becoming ubiquitous. AI-based systems to identify, optimize, automate, and scale solutions to complex economic and societal problems are being proposed and implemented. This has motivated regulation efforts, including the Proposal of an EU AI Act. This interdisciplinary position paper considers various concerns surrounding fairness and discrimination in AI, and discusses how AI regulations address them, focusing on (but not limited to) the Proposal. We first look at AI and fairness through the lenses of law, (AI) industry, sociotechnology, and (moral) philosophy, and present various perspectives. Then, we map these perspectives along three axes of interests: (i) Standardization vs. Localization, (ii) Utilitarianism vs. Egalitarianism, and (iii) Consequential vs. Deontological ethics which leads us to identify a pattern of common arguments and tensions between these axes. Positioning the discussion within the axes of interest and with a focus on reconciling the key tensions, we identify and propose the roles AI Regulation should take to make the endeavor of the AI Act a success in terms of AI fairness concerns.

Eleven Years of Gender Data Visualization: A Step Towards More Inclusive Gender Representation

“…The purpose of this action is to safeguard against discrimination based on these attributes. For example, the use of protected attributes in AI is strictly regulated [17]. These attributes deserve careful and thorough attention from the visualization community, and others have already started the discussion.…”

Operationalizing Explainable Artificial Intelligence in the European Union Regulatory Ecosystem

“…As for the GDPR, objections to fully interpretable design criteria find appeal in the EU directive establishing trade secret integrity and IP rights. While transparency is desirable to respect EU user rights, it is problematic to unilaterally maximize an AI system transparency when the same users are not controlled over possible misuses [17], [18]. In this regard, end-user liability can be ascertained from the need, expressed in Art.…”

An Operational Perspective to Fairness Interventions: Where and How to Intervene

“…Several recent works have drawn attention to the complexities of fairness in ML beyond the algorithms. These works bring in interdisciplinary perspectives on fairness, including arguments from the legal [15,6,60], economic [55,1], moral [42,15], and industrial [51,7,4,2] perspectives. These works highlight the need to concretely evaluate how group data is used for fairness purposes and to what extent interventions work without such data.…”