Automating access to social assistance benefits (ASA)

Automated Decision Making (ADM) is increasingly deployed in modern welfare states to provide services and allocate benefits. It is argued that ADM systems know no prejudice, ensure equal treatment, are better equipped to identify potential beneficiaries, and are cost-effective. In contrast, ADM systems can also compound the effects of simple errors, can be biased and perpetuate prejudice while grave concern was voiced about a lack of transparency. Research on the outcomes of ADM in the field of social policy is long overdue. In this project we investigate whether and if so, to what extent and under which conditions, ADM leads to equal, accurate and transparent decisions in the Belgian social assistance system. We evaluate current practices, we develop new ADM tools to automate and improve the social assistance benefit application procedure, and we investigate how this affects professionals’ and the relationship with their clients, all with a focus on the outcomes for vulnerable clients.

Keywords:decision-making, automation, application tool, social assistance, algorithms

Disciplines:Counselling, welfare and community services, Social policy, Public policy, Research, science and technology policy