Mathematician | Interdisciplinary PhD candidate

Research at the intersection of Mathematics & Legal Studies & Science and Technology Studies

Email

me [at] paolalopez [dot] eu

Affiliations

  • University of Vienna: click
  • The Privacy & Sustainable Computing Lab: click
  • Academy of Fine Arts Vienna: click

Upcoming Publication

Lopez, Paola (2019): Reinforcing Intersectional Inequality via the AMS Algorithm in Austria, Conference Proceedings of the STS Conference Graz 2019, 6th – 7th May 2019, DOI: 10.3217/978-3-85125-668-0-16

Preprint available here:

Abstract. This paper examines the so-called AMS Algorithm from a mathematical perspective for a non-mathematical audience: this algorithmic system constitutes a predictive model that will be used by the Public Employment Service Austria (AMS) starting in 2020 to algorithmically classify job-seekers into three groups, each with different access to AMS support resources, according to their predicted chances on the labour market. Since the features gender, age, childcare responsibilities, disability and citizenship are explicitly implemented in the model and are thus linked to the availability of resources, this algorithmic system is to be considered very problematic. This paper is part of an ongoing research project, and it identifies three conceptual building blocks of the AMS Algorithm that are all based on human decisions and in which obvious societal bias can be located. Furthermore, this model is used as an illustrative example to address the larger question of what can be expected when predictions are made that are based solely on data that describes the past: If the predictions by these models result in unquestioned and confirmatory measures such as the redistribution of resources, a reproduction and reinforcement of inequality is possible. If these measures are now applied to vulnerable and highly dependent target groups, such as job-seekers, it will be more drastic: In a first step, these predictive models depict the reality of discrimination, then, in a second step, normatively reinforce it as a supposedly objective fact and finally, in a third step, return it to the social sphere by means of the resulting measures.

Media

  • Netzpolitik.org, October 10, 2019 [German] click
  • AlgorithmWatch.org, October 6, 2019 [English] click
  • Futurezone.at, October 3, 2019 [German] click
  • Kurier, September 28, 2019 in print [German] click
  • Futurezone.at, September 27, 2019 [German] click