Seite - 39 - in Austrian Law Journal, Band 1/2019

ALJ 2019 Digital Single Market – towards Smart Regulations 39 Although the topics discussed are very diverse, one common link between them must not be overlooked: The suitable legal basis for the creation of a Digital Single Market will in most cases be provided by Art. 114 of the Treaty on the Functioning of the European Union (TFEU).8 II. Machine-learning algorithms A. Definition, types and use cases The first challenge for the Digital Single Market discussed in this paper is the increased use of autonomous decision making by machine-learning algorithms in all aspects of life. In contrast to conventional algorithms with a specified rule-based behaviour, these machine-learning algorithms are able to learn independently and act autonomously.9 From a legal perspective, the handling of machine-learning algorithms and their ability of making increasingly autonomous decisions is still largely unclear. So far, there is no general codification concerning (autonomous) algorithms,10 legislation which refers to them or regulates their use only exists to a very limited extent (see also sub-section 2.2. below).11 A major challenge in creating comprehensive legislation on algorithms is the rapid pace of technological development, which entails the risk of creating regulations that are outdated before even having entered into force. The difficulty of grasping this phenomenon starts with the fact that there is no uniform definition of machine-learning algorithms which could form the basis of a legal regulation. The exact meaning of the very broad term “algorithm” depends on the context in which it is used.12 Machine-learning algorithms are now in use throughout our life.13 In contrast to deterministic algorithms, machine-learning algorithms learn autonomously based on a given (training) data set. This is to some extent comparable to the functioning of the human brain which “naturally” uses 8 Cf. also Korte, Art. 114 AEUV, in EUV/AEUV, para 161 (Calliess and Ruffert ed., 5th ed. 2016). However, the limits set out in the jurisdiction of the CJEU (cf. in particular Case C-376/98 Federal Republic of Germany v European Parliament and Council of the European Union, ECLI:EU:C:2000:544 and Case C-380/03 Federal Republic of Germany v European Parliament and Council of the European Union, ECLI:EU:C:2006:772) must be observed. 9 Cf. RUSSELL AND NORVIG, ARTIFICIAL INTELLIGENCE: A MODERN APPROACH (3rd ed. global edition 2016). 10 Nevertheless, the question how AI could and should be regulated is already being widely discussed within the legal community. Reed, How should we regulate artificial intelligence?, 376 PHILOSOPHICAL TRANSACTIONS OF THE ROYAL SOCIETY A (2018), https://doi.org/10.1098/rsta.2017.0360; Scherer, Regulating Artificial Intelligence Systems: Risks, Challenges, Competencies, and Strategies, 29 HARVARD JOURNAL OF LAW & TECHNOLOGY 354 (2016); Yeung, Algorithmic Regulation: A Critical Interrogation, 12 REGULATION & GOVERNANCE 505 (2018). 11 Algorithms are mentioned explicitly in Sec 14 of the Austrian regulation on gambling machines (Automatenglücksspielverordnung). This provision refers to the use of an algorithm in the context of the determination of game results. Further examples in the Austrian legal system are Sec 11 of the Stock Exchange Act 2018 (BörseG 2018) and Sec 12 of the Federal Equal Treatment Act (B-GlBG). The German law professor von Lewinski recently attempted to arrange the punctually existing regulations of algorithmic law in the form of a matrix with the aim of bringing order into a new field of law (cf. von Lewinski, Regulierungsbedarf und Regulierungsfelder von algorithmischen Systemen, ZEITSCHRIFT FÜR INNOVATIONS- UND TECHNIKRECHT 168 [2018]). 12 Cf. different perspectives in Mohabbat Kar, Thapa and Parycek (ed.), (Un)berechenbar? Algorithmen und Automatisierung in Staat und Gesellschaft (2018) Fraunhofer-Institut für Offene Kommunikationssysteme FOKUS, Kompetenzzentrum Öffentliche IT https://nbn-resolving.org/urn:nbn:de:0168-ssoar-57518-2. 13 Cf. Makridakis, The forthcoming Artificial Intelligence (AI) revolution: Its impact on society and firms, 90 FUTURES 46 (2017).