Click the star to add/remove an item to/from your individual schedule.
You need to be logged in to avail of this functionality.
Log in
- Convenors:
-
Doris Allhutter
(Austrian Academy of Sciences)
Karolina Sztandar-Sztanderska (Institute of Philosophy and Sociology, Polish Academy of Sciences)
Astrid Mager (Austrian Academy of Sciences)
Send message to Convenors
- Chair:
-
Doris Allhutter
(Austrian Academy of Sciences)
- Format:
- Traditional Open Panel
- Location:
- HG-01A33
- Sessions:
- Friday 19 July, -, -
Time zone: Europe/Amsterdam
Short Abstract:
The datafication of public welfare has far reaching implications as regards social inclusion, equality, and democracy. This panel asks how welfare is automated across various domains and how data-based infrastructures in public administration co-emerge with a transformation of the welfare state.
Long Abstract:
In recent years, the welfare sector has been facing increasing demands and shrinking resources across the EU and beyond. Calls to mitigate social hardships more effectively and increase public administration's efficiency suggest introducing data-driven decision-support and enhancement with artificial intelligence. The datafication of core welfare services such as employment services, healthcare and social benefits provision has far reaching implications as regards social inclusion, equality, governance, and democracy. Data-based infrastructures in public administration have a significant impact on the living circumstances of citizens and on human agency. The notion of ‘infrastructures of welfare’ suggests that the implementation of automated systems is closely entangled with a transformation of administrative work practices and of the relationship between citizens and the state.
This panel addresses the digital transformation of public welfare and invites scholars from science and technology studies, critical data studies, political science, sociology, organization studies, administrative science, computer science and beyond for contributions on algorithmic welfare from an interdisciplinary perspective. Contributions may address the following questions and more:
How is welfare automated across Europe and beyond? How is automated decision-making in the public sector co-produced with a transformation of welfare? How do datafication and automation change power relations within the welfare state (e.g., between the public and private sector, between various levels of government)? How do welfare infrastructures and automation affect administrative practices and discretion? How is the relationship between citizens and the state reconstituted with emerging welfare infrastructures? How can the algorithmic governance enacted by welfare infrastructures be theorized?
Accepted papers:
Session 1 Friday 19 July, 2024, -Short abstract:
This contribution presents the use and limitations of the notion of “tool of government” (Desrosières, 2008) to study algorithms together with their administrative contexts. It highlights its generativity by examining the case of the French risk-scoring algorithm to combat “social fraud”.
Long abstract:
Stuck at a disciplinary crossroads, the study of algorithms in administrative contexts often focuses either on singular technological artefacts or on broad governmental logics at play. The risk scoring algorithm to combat “social fraud” in the French Social security system is no exception. While criticism and mobilisation concentrated on the algorithm’s reliance on discriminatory variables, on the effects of automated decision-making and its lack of transparency (la Quadrature du Net, 2022; Geiger et al., 2023), sociologist Vincent Dubois highlighted the importance of contextualising the algorithm within the history of control policy, its practices, and varying conceptions of “social fraud” (Dubois 2012; 2022).
Studying algorithms as tools of government offers a middle ground for the study of technology within administrative contexts (Le Galès and Lascoumes, 2004), provided it uses a relational and materialist conception of technological objects (Simondon, 2012). In our case, it allows to follow the algorithm's genesis, how its production and deployment relied on the adoption of an increasingly individualized conception of risk (Dubois), on vast data collection and data sharing frameworks across administrations (Poulain, 2022). Its usage required to insert risk-scores with administrative regimes of proof and to reorganize pre-existing practices of control.
Examining jointly the co-construction of administrative fields and their tooling, such an approach allows to refine and ground materially sociological-political literature on welfare reform and privatization (Da Silva, 2022) and to explore the ways in which choices of administrative tools take part in policy orientations within established fields of social action.
Short abstract:
Data infrastructures are a core component of understanding social change. Thus, the paper focuses on a case study in a primary care unit, exploring the new territorial organization that becomes crucial for the construction of a public health data infrastructure that allows personalized interventions
Long abstract:
Data infrastructures are transforming state-citizen relations through the logic of personalized risk and the individualization of social problems. This paper focuses on a case study in a primary care unit in Italy called Community House, which started in 2023 and is ongoing as part of a larger research project AUTOWELF. The aim of the paper is twofold: to explore the transformation of the local dimension of territorial care with the introduction of a new concept of community that recognizes the values of social ties and enhances relational resources, while simultaneously, new territorial organization becomes crucial for the construction of a public health data-based infrastructure that allows personalized interventions. This mode of framing the local level is assisted by the new use of technology and ICT systems, which are closely linked and play a key role in enhancing the managerial logic and strategies in shaping public administration and the reorganization of healthcare. This contributes to a significant change in universal healthcare interventions and is connected to the concept of the datafication of health and "personalized medicine", which focus on the individual as a precise standard. Based on a qualitative study of document analysis and semi-structured interviews, the analysis will focus on the development of a decision support system for the Chronic Care Model as an attempt to shed light on the experiences and challenges of how algorithms that are implemented for the decision-making process are both contributing to changing the type of bureaucracy and the discretionary capabilities of professionals.
Short abstract:
This paper critically examines a fraud detection software used by an Austrian health insurance to detect bogus companies. What drives the software, what knowledge is created, and how that relates to wider transformations of the welfare state will be discussed.
Long abstract:
All over Europe, algorithmic systems are introduced to detect welfare fraud more efficiently. Within a wider context of new managerialism, these software tools conduct risk scoring and identify irregular patterns in large data-sets. This implies the incorporation of typical implications of data analytics –surveillance, bias, discrimination – into public sectors (Dubois et al. 2018). Moreover, social practices of welfare institutions change due to the implementation of algorithmic systems and the numeric “evidence” they create that case workers have to balance with their own “instincts” (Allhutter et al. 2021).
This paper critically examines a fraud detection software used by an Austrian health insurance to detect bogus companies (FWF I 6075). In Austria, a law has been passed in 2015 to fight welfare fraud and make its detection mandatory for public institutions. Accordingly, the software repurposes administrative data to conduct data forensics and help the financial police. Based on qualitative interviews with software developers, case workers, and representatives of the health insurance, our analysis will focus on the following questions: What drives the development of the software? What knowledge, or “evidence”, is created with the software and how is it framed in the context of fraud detection? How do larger socio-political transformations of the welfare state impact both practices and narratives? Theoretically, the paper draws on STS and knowledge production, critical algorithm studies, and public policy literature. To conclude, we discuss larger questions related to the automation of the welfare state and its impact on social practices, social roles, and social inequalities.
Short abstract:
This paper explores the roles regulatory discourses allocate to infrastructures and data in the organisation of digital welfare. Drawing on a data feminist analysis, I discuss what power relations between human and more-than-human actors within the public sector these discourses hence configure.
Long abstract:
With the ongoing datafication and automation of welfare services provision, digital data are used for governance, public decision-making, (non-)citizens’ profiling and scoring for e.g. fraud prediction or calculation of welfare benefits. In regulatory discourses (e.g. strategy papers) datafication and automation of public welfare are, hence, promise ‘solutions’ to certain problems. This contribution joins scholarship critically interrogating such promises by exploring discourses around datafication and (potential) automation of the public sector, particularly in Germany, more closely.
I ask, what roles regulatory discourses allocate to infrastructures and data in the organisation of digital welfare. Drawing on a data feminist approach (D’Ignazio & Klein 2020) and research on imaginaries and metaphors (Mager & Katzenbach, 2021), I discuss what power relations between human and more-than-human actors within the public sector these discourses hence configure. This analytical lens helps interrogate public welfare actors’ discourses in regard to their values, whose work would be required to fulfil the promises of datafication or (potential) automation of welfare, and who are the expected beneficiaries. Empirically, the paper is based on a discourse analysis of publicly available documentation regulating the datafication and automation of German public sector (strategic papers, legal acts, policy documents). By attending to these discourses and speculating about subsequent changing relations between the state and its citizens through the feminist analysis of data power, this contribution discusses the extent to which datafication and automation could contribute to more just and equitable futures and what (other) relations are required to achieve these.
Short abstract:
Departing from the perspective of people involved in the Lisbon Intelligent Management Platform and official state narratives, the paper reflects on expectations of the deployment of ADM in urban data management and the promotion of collective welfare at the city level.
Long abstract:
The Lisbon Intelligent Management Platform (PGIL) gathers databases from different origins, providing the Lisbon City Council and other public services dashboards to work with. PGIL, therefore, is a data integration platform that supports activities performed by civil defense, fire brigade, and public security officials. As part of the AUTO-WELF project, the platform has been explored as a case study to understand data analysis and automation deployment to promote collective welfare. In this context, the paper combines findings of the interviews carried out in September-October 2023 with the platform team and official document analysis to address views and expectations concerning Automated Decision-Making (ADM) in urban data management, exemplifying how infrastructures of welfare may be materialized in a city of a Southern European country. To conduct such an analysis, the paper mobilizes Feminist STS and critical data studies, in addition to the “right to the city” concept, with the purpose of contributing to the discussion around a scenario that has multiplied throughout Europe. Although ADM is not deployed in PGIL, narratives on the possibility of using it emerged in the interviews and documents, as an expectation desired for the platform’s future or a next step to be accomplished. Thus, departing from the perspective of people involved in the development of systems of this kind and official state narratives, the empirical information addressed in the paper helps us think about algorithms and welfare at the city level, also considering citizens’ needs and priorities.
Short abstract:
Based on a case study of the profiling algorithm labour market policies, the article addresses the politics of algorithmic articulation of law. More specifically, we analyze how the law is articulated by the computer means, what discrepancies between the law and ADM are produced, by whom, and how.
Long abstract:
Understanding how the law is articulated by the computer means is important in the context of increasing use of algorithmic decision-making systems (ADM) in public policies and high profile scandals (e.g. Angwin et al. 2016, van Bekkum & Borgesius 2021; Rachovitsa & Johann 2022). Based on a case study of the profiling algorithm deployed in Poland to differentiate unemployed persons’ obligations and access to active labor market programs, the article contributes to the debates on computer representation and translation of law (Bovens, Zouridis 2002; Zouridis et al 2020; Hilderbrandt 2013, 2018). It addresses the questions of how the law is articulated by the computer means, what discrepancies between the law and ADM are produced, by whom, and how. Using unique data concerning the algorithm and its development, we demonstrate important discrepancies between the legal and algorithmic framework that are indicative of backstage discretionary decision-making and go far beyond what is assumed by the literature as a necessary by-product of representation or translation of law. Second, we reconstruct how these discrepancies came into being to conclude that surprisingly the discretion was mainly exercised by old policy-makers, namely the representatives of executive power and public administration, rather than by statisticians, data analysts, programmers or other so-called ‘system-level bureaucrats’ (Bovens & Zouridis, 2002) or ‘back-office policy-makers’ (Ustek-Spilda 2020) as the literature would suggest.