Regulating Algorithmic Decision Systems
0
조회
0
다운로드
1
인용
0
참고문헌
키워드
초록
Automated decision-making plays an increasing role in life. The author identifies various starting points for regulation, such as the three components of the knowledge triangle: data, algorithms and the generated knowledge. Considering that technologies are always integrated into practices, a differentiated scope of regulation is inevitable and starting points for regulation can be enlarged. Other options for regulation can be the technical level, the context of the application or the resulting business models. Because of a lack of examination of whether a problem is caused due to its digital practice or roots in the corresponding analog practice, he proposes analyzing and comparing the analog practices with the digital practice to figure out an appropriate legal framework. Afterward, he outlines that ADS is already incorporated in Art. 22 GDPR, Section 35a of the German Administrative Procedure Act and Section 155 of the German Fiscal Code. After discussing various objections against ADS, for example, the question of correlation and causality, the problem of discrimination or lack of social competence, he notes that the derogations of Art. 22 GDPR and Section Section 35a of the German Administrative Procedure Act assume human decision to be superior. He doubts this generalization and says that instead of banning such instruments, it is more important to develop accompanying safeguards for automatic decision-making processes. One application for ADS can be predictive analytics, an already partly regulated technology to prevent or reduce crimes using nonpersonal data and patterns. Contrary to widespread concern, predictive policing can prevent nonspecific investigations and should be regulated, not prohibited. A positive example for regulation of DSS is the European legal framework for Clinical Decision Support Systems. They aim at enhancing medical decisions with clinical knowledge and are regulated by a differentiated risk-based approach. In the end, the author deals with future regulation measures for ADS/DSS, such as certification systems and conformity assessments. The recent proposal of the EU for a legal framework on AI incorporates both and establishes a risk-based approach, which is an important step to a more sophisticated regulation.
Abstract
Automated decision-making plays an increasing role in life. The author identifies various starting points for regulation, such as the three components of the knowledge triangle: data, algorithms and the generated knowledge. Considering that technologies are always integrated into practices, a differentiated scope of regulation is inevitable and starting points for regulation can be enlarged. Other options for regulation can be the technical level, the context of the application or the resulting business models. Because of a lack of examination of whether a problem is caused due to its digital practice or roots in the corresponding analog practice, he proposes analyzing and comparing the analog practices with the digital practice to figure out an appropriate legal framework. Afterward, he outlines that ADS is already incorporated in Art. 22 GDPR, Section 35a of the German Administrative Procedure Act and Section 155 of the German Fiscal Code. After discussing various objections against ADS, for example, the question of correlation and causality, the problem of discrimination or lack of social competence, he notes that the derogations of Art. 22 GDPR and Section Section 35a of the German Administrative Procedure Act assume human decision to be superior. He doubts this generalization and says that instead of banning such instruments, it is more important to develop accompanying safeguards for automatic decision-making processes. One application for ADS can be predictive analytics, an already partly regulated technology to prevent or reduce crimes using nonpersonal data and patterns. Contrary to widespread concern, predictive policing can prevent nonspecific investigations and should be regulated, not prohibited. A positive example for regulation of DSS is the European legal framework for Clinical Decision Support Systems. They aim at enhancing medical decisions with clinical knowledge and are regulated by a differentiated risk-based approach. In the end, the author deals with future regulation measures for ADS/DSS, such as certification systems and conformity assessments. The recent proposal of the EU for a legal framework on AI incorporates both and establishes a risk-based approach, which is an important step to a more sophisticated regulation.
참고문헌 (42)
- "Abgeordnetenhaus Berlin: Drucksache 18/17562 v. 14" (2019)[URL]
- Aden, Hartmut, "Defizite der Polizeirechtsentwicklung und Techniknutzung", Zeitschrift für Rechtspolitik ZRP, 52(6), pp.175~ (2019)
- Akerlof, George A, "How human psychology drives the economy, and why it matters for global capitalism" (2009)
- Arzt, Clemens, "Das neue Gesetz zur Fluggastdatenspeicherung", Die öffentliche Verwaltung DÖV, 70(24), pp.1023~ (2017)
- Berner, Eta S, "Clinical Decision Support Systems", pp.1~ (2016)
- BDSG Datenschutz-Grundverordnung, "DS-GVO", pp.2020, Buchner, Benedikt
- DeLamater, John, "Social Psychology", ~ (1998)
- "Regulation on a European Approach for Artificial Intelligence – Draft" (2021)[URL]
- Fiske, Susan T, "Handbook of Social Psychology", pp.357~ (1998)
- Gerstner, Dominik, "Predictive Policing in the Context of Residental Burglary: An Emirical Illustration on the Basis of a Pilot Project in Baden-Württemberg, Germany", European Journal for Security Research, 3(2), pp.115~ (2018)
- Gless, Sabine, "Rechtsstaatlicher Strafprozess und Bürgerrechte", pp.165~ (2016), GS Weßlau
- "Zertifizierung von KI-Systemen, White paper aus der Plattform Lernende Systeme" (2020)[URL]
- Hornung, Gerrit, "Verwaltungsrecht VwVfG Kommentar" (2020)
- Kainz, Viola, "Fluggastdaten im Kampf gegen schwere Kriminalität, Der Weg von der PNR-Richtlinie ins nationale Recht", SIAK-Journal für Polizeiwissenschaft und polizeiliche Praxis, 14(4), pp.74~ (2018)
- Kaufmann, Mareile, "Predictive Policing and the Politics of Patterns", The British Journal of Criminology, 59(3), pp.674~ (2019)
- "Data Science, 2018. Knobloch, Tobias: Vor die Lage kommen: Predictive Policing in Deutschland" (2018)[URL]
- Kostov, Iva, "Intelligent Technologies and Applications 2020", pp.392~ (2021)
- Kux, Benjamin R, "Einflussfaktoren für die Implementierung, Verbreitung und Verstetigung von Clinical Decision Support Systems", ~ (2018)
- "Predictive Policing – eine Bestandsaufnahme" (2014)[URL]
- "Abschluss bericht SKALA" (2018)[URL]
- "Predicitive Policing in der Schweiz:Chancen, Herausforderungen, Risiken, Bulletin zur Schweize rischen Sicherheitspolitik" (2018)[URL]
- Lepri, Bruno, "Transparent and Accountable Algorithmic Decision-making Processes: The Premise, the Proposed Solutions, and the Open Challenges", Philosophy & Technology, 31(3), pp.611~ (2018)
- "Das metrische Wir, 2017. Müller-Quade, Jörn et al.: Sichere KI-Systeme für die Medizin, Whitepaper aus der Plattform Lernende Systeme" (2020)[URL]
- Nassehi, Armin, "Muster", ~ (2019)
- "Predictive Policing. The Role of Crime Forecasting in Law Enforcement Operations" (2013)[URL]
- "Kriminologische Forschungsstelle, Prädiktionspotential schwere Einbruchskriminalität - Ergebnisse einer wissenschaftlichen Befassung mit Predictive Policing" (2019)[URL]
- Rademacher, Timo, "Predictive Policing im deutschen Polizeirecht", Archiv des öffentlichen Rechts AöR, 142(3), pp.366~ (2017)
- Rätke, Bernd, "Abgabenordnung" (2020), Kommentar
- "Predictive Policing: Beobachtungen und Reflexionen. Zur Einführung und Etablierung einer vorhersagenden Polizeiarbeit" (2016)[URL]
- Rüsken, Reinhart, "Abgabenordnung Kommentar" (2020)
- Schäfer, Hilmar, "Der Gebrauch des Digitalen", Zur praxeologischen Analyse digitaler Kultur, 30(1) (2021)
- Scholz, Philip, "Datenschutzrecht" (2019)
- Schwarz, Norbert, "Sozialpsychologie", pp.355~ (1997)
- "Predi ctive Policing in Germany, Konstanzer Online -Publikationssystem" (2018)[URL]
- Siegel, Erika H, "Lisa: Emotion fingerprints or emotion populations? A meta-analytic investigation of automatic features of emotion categories", Psychological Bulletin, 144(4), pp.343~ (2018)
- Singelnstein, Tobias, "Predictive Policing:Algorithmenbasierte Straftatprognosen zur vorausschauenden Kriminalintervention", Neue Zeitschrift für Strafrecht NStZ, 38(1), pp.1~ (2018)
- "An overview of clinical decision support systems: benefits, risks, and strategies for success" (2020)[URL]
- Trute, Hans-Heinrich, "Der Stellenwert von Real-World Evidenz im deutschen Sozial- und Gesundheitsrecht", Gefasschirurgie, 24(3), pp.228~ (2019)
- Trute, Hans-Heinrich, "Digitalisierung im Spannungsfeld von Politik", pp.313~ (2018), Wirtschaft und Recht
- Ulbricht, Lena, "When Big Data Meet Securitization. Algorithmic Regulation with Passenger Name Records", European Journal for Security Research, 3(2), pp.139~ (2018)
- Wasylewicz, Arthur T. M, "Fundamentals of Clinical Data Science", pp.153~ (2019)
- Zuiderveen Borgesius, Frederik J, "Strengthening Legal Protection against Discrimination by Algorithms and Artificial Intelligence", The International Journal of Human Rights, 24(10), pp.1572~ (2020)