Download PDFOpen PDF in browser

Algorithmic Profiling in Public Employment Services: a Systematic Review on the Effects on Caseworkers and Jobseekers

EasyChair Preprint 13691

32 pagesDate: June 17, 2024

Abstract

For decades, Public Employment Services (PES) have used algorithmic systems to profile jobseekers. Profiling scores assist caseworkers in selecting appropriate labor market policy instruments for jobseekers. While statistical profiling is currently the leading method, some countries are starting to implement artificial intelligence (AI), specifically models that make predictions about jobseekers (Desiere et al. 2021). These models use statistical or supervised machine learning algorithms to assist caseworkers in making decisions (Kleinberg et al., 2018). All algorithmic profiling systems aim to improve the efficiency of the PES. However, some systems have faced criticism from academia and civil rights organizations. This criticism is centered around the difficulty jobseekers face in rejecting an assigned score and the limited discretion caseworkers have due to system design. This study investigates the impact of algorithmic profiling on caseworkers and jobseekers. Specifically, we examine (1) the level of discretion caseworkers have when using algorithmic profiling, (2) the options available to jobseekers to challenge the results of algorithmic profiling, (3) the transparency of these systems, and (4) the impact of algorithmic profiling scores on resource allocation. We conducted a qualitative systematic literature review to identify relevant articles. We developed a coding scheme to analyze the articles based on our research questions. This study enhances the understanding of the impact of digital transformation on the work of street-level bureaucrats. The paper contributes to the public debate by systematically identifying and compiling a differentiated picture of the impact and risks of algorithmic profiling.

Keyphrases: Artificial Intelligence, Public Employment Services, Systematic Literature Review, algorithmic profiling, street-level bureaucrats

BibTeX entry
BibTeX does not have the right entry for preprints. This is a hack for producing the correct reference:
@booklet{EasyChair:13691,
  author    = {Mareike Sirman-Winkler and Daria Szafran and Sonja Mei Wang},
  title     = {Algorithmic Profiling in Public Employment Services: a Systematic Review on the Effects on Caseworkers and Jobseekers},
  howpublished = {EasyChair Preprint 13691},
  year      = {EasyChair, 2024}}
Download PDFOpen PDF in browser