How Should We Regulate AI? Current Rules and Principles as Basis for 'Responsible Artificial Intelligence'

Abstract

This paper is a revised, updated and extended version of my paper published in Ordnung der Wissenschaft (OdW) 1, 2020, 9-21. It will be published 2020 in a book edited by Hanno Merkt (Universitat Freiburg). One challenge of the 21st century to humankind, democratic and non-democratic States, is the widespread use of Artificial Intelligence (AI). Hardly any day passes without news about the disruptive force of AI – both good and bad. Some warn that AI could be the worst event in the history of our civilization. Others stress the chances of AI diagnosing and fighting diseases as cancer or COVID-19, or supporting humans in the form of autonomous cars. But because AI is disruptive, the call for its regulation is widespread, including the call by some actors for international treaties banning so-called “killer robots”. Nevertheless, until now, there is no consensus on how and to what extent we should regulate AI. This paper examines whether we can identify key elements of responsible AI, spells out what exists as part of “top-down” regulation, and how new guidelines, such as the 2019 OECD Recommendations on AI can be part of a solution to regulate AI systems. In the end, I propose a solution that is coherent with international human rights law in order to frame the challenges posed by AI that lie ahead of us without undermining science and innovation; reasons are given why and how a human rights-based approach to responsible AI should inspire new rules at the international level.