Paper/Speech Details of Conference Program for the 31st NISPAcee Annual Conference Program Overview WG9: The Rule of Law & Public Administration Author(s) Matej Babsek University of Ljubljana Ljubljana Slovenia Polonca Kovač, University of Ljubljana, Faculty of Public Administration, Ljubljana, Slovenia Title Social Procedures and Artificial Intelligence: Opportunities and Barriers in Slovenia and Beyond File Paper files are available only for conference participants, please login first. Presenter Matej Babsek Abstract The unpredictable and complex social conditions of nowadays accelerate the digital transformation of public administration which results in introduction of innovative artificial intelligence mechanisms into conducting administrative procedures. In social procedures, automated-decision making and the use of algorithms for detecting welfare frauds ensure efficient, fair and transparent redistribution of public funds while pursuing the constitutional principle of the rule of law shielding vulnerable population from arbitrary actions of the authorities. However, for automation to become legitimate, (administrative) empathy as the ability to understand parties’ life-situations from multiple perspectives, is crucial, especially due to the specific nature of social procedures. In Slovenia, a country with 2 mio. inhabitants, there are as many as approximately 1.8 mio. social procedures each year, predominantly conducted by social work centres (SWC). Every automation using artificial intelligence in these procedures therefore significantly reduces red tape for both, parties and authorities. From the perspective of good governance principles, the rationale for this study is to identify opportunities and possible drawbacks of introducing artificial intelligence mechanisms into social procedures in Slovenia. Our research question is which solutions of artificial intelligence, given the specifics of this environment, could be introduced into the social procedures in Slovenia, and which measures for their implementation should be ensured prior. Initially, some fraud-flagging welfare projects (e.g., MiDAS in the US for detecting unemployment fraud, child benefits fraud algorithm in the Netherlands, etc.) and selected automated decision-making welfare solutions will be critically analysed. Next, a focus group with five Slovenian NGOs as representatives of their client users as parties in social procedures led by SWCs will be conducted, focusing on empathy as precondition for legitimate and fair decisions in automated procedures. Secondary analysis of statistical data on two existing Slovenian informational calculations – one on social rights and other on pre-filled informational calculation of personal income tax, will be carried out next. Finally, the main aspect of digitalization in social procedures is how to do it in a way that minimizes harms for most vulnerable parties. In this regard, SWOT analysis will be carried out, highlighting pros and cons of using artificial intelligence in social procedures, proposing much needed trade-off measures in the further development of social procedures in Slovenia and in other CEE countries.
