Responsible Algorithmization in the Public Sector: A Natural Perspective Based on Ethnographic Research in Regulation, Policing, and Healthcare

• 成果类型: Article; Early Access
• 署名作者: Meijer, Albert; Lorenz, Lukas; Donatz-Fest, Isabelle; Hoff, Jan-Luuk
• 署名单位: Utrecht University; Vrije Universiteit Amsterdam
• 刊物名称: PUBLIC ADMINISTRATION
• ISSN/ISSBN: 0033-3298; 1467-9299
• DOI: 10.1111/padm.70092
• 发表日期: 2026-09-11
• 关键词: artificial-intelligence; Complexity theory; management; POLITICS; ACCOUNTABILITY; TECHNOLOGY

摘要

Responsible algorithmization refers to the use of algorithms in the public sector while safeguarding public values. Approaches to responsible algorithmization suggested by the EU, OECD, and others depart from a rational perspective and highlight formal structures and procedures. This paper complements this perspective by adding a natural perspective, which emphasizes sense-making and actual practices. This natural perspective is used for a set of ethnographic studies of algorithmization in three domains in the Netherlands: regulation, policing, and healthcare. Our findings show that algorithmization of government organizations is an emergent, political, open, and fragmented process, rather than, as assumed by rational approaches to responsible algorithmization, pre-structured, neutral, closed, and simple. This means that rational approaches fall short of providing a comprehensive basis for establishing responsible algorithmization. We conclude that not only passive responsibility but also active responsibility for algorithmization in the public sector needs to be addressed to safeguard public values. La algoritmizaci & oacute;n responsable se refiere al uso de algoritmos en el sector p & uacute;blico salvaguardando los valores p & uacute;blicos. Los enfoques sobre algoritmizaci & oacute;n responsable sugeridos por la UE, la OCDE y otros organismos parten de una perspectiva racional y destacan las estructuras y los procedimientos formales. Este art & iacute;culo complementa dicha perspectiva al a & ntilde;adir una perspectiva natural, la cual enfatiza la atribuci & oacute;n de sentido y las pr & aacute;cticas reales. Esta perspectiva natural se utiliza en un conjunto de estudios etnogr & aacute;ficos sobre la algoritmizaci & oacute;n en tres dominios en los Pa & iacute;ses Bajos: la regulaci & oacute;n, la polic & iacute;a y la sanidad. Nuestros hallazgos muestran que la algoritmizaci & oacute;n de las organizaciones gubernamentales es un proceso emergente, pol & iacute;tico, abierto y fragmentado, en lugar de ser preestructurado, neutral, cerrado y simple, como asumen los enfoques racionales de la algoritmizaci & oacute;n responsable. Esto significa que los enfoques racionales se quedan cortos a la hora de proporcionar una base integral para establecer una algoritmizaci & oacute;n responsable. Concluimos que, para salvaguardar los valores p & uacute;blicos, es necesario abordar no solo la responsabilidad pasiva sino tambi & eacute;n la responsabilidad activa en la algoritmizaci & oacute;n del sector p & uacute;blico.