The Use of Machine Learning in Public Organizations - an Interview with PhD Student Friso Selten
Friso Selten recently started a PhD position that is part of the SAILS program. This PhD project is a collaboration between FGGA, LIACS, and eLaw, and is supervised by Bram Klievink (FGGA), Joost Broekens (LIACS), and Francien Deschene (eLaw). In the project Friso will investigate the influence of artificial intelligence within public organizations with a specific focus on decision-making processes. ‘I became particularly enthusiastic about the topic of AI and its application in the public sector while studying for a Master's degree in Data Science at the University of Amsterdam. This inspired me to follow the research master's in Public Administration and Organisational Science at Utrecht University. In this program, I primarily focused on the impact that algorithms have within the street-level bureaucracy. I hope to further investigate these and other aspects of the connection between AI, public administration, and ethics in this PhD project.’
What is your research focused on?
‘My research focusses on the use of machine learning within public organizations. Machine learning approaches hold the promise to make governance more efficient, flexible, and responsive. When implemented in the public sector, machine learning applications also become subject to critique. Algorithms are feared to introduce bias and unfairness in decision-making. The adverse effects of machine learning applications are, to some extent, caused by these technologies being a ‘black box’. Machine learning algorithms follow complex mathematical rules that are not understandable to a human. Controlling and correcting algorithmic procedures is, therefore, a complex task. Many researchers are working on making machine learning algorithms both accurate and transparent.’
‘However, the technical opaqueness of algorithms is not the only aspect that hinders the correct implementation and use of these technologies in public organizations. While many studies investigated the outcomes of algorithms within public organizations, the innovation and implementation process itself has typically also been treated as a black box. Important questions that will be answered in this research are: how can algorithms best be developed, supported, and sustained within public organizations, and what managerial, organizational, cooperative, and executive capabilities do public organizations need to incorporate the advantages that algorithms have and reduce adverse effects?’
'While many studies investigated the outcomes of algorithms within public organizations, the innovation and implementation process itself has typically also been treated as a black box.'
What is the goal you want to reach with this research?
‘The goal of this project is to provide public organizations with the necessary tools to use machine learning approaches to create public value. Public value describes the ensemble of values that public organizations have to provide, such as efficiency and effectiveness, fairness and transparency, and robustness and adaptiveness. I will study the public values organizations aim to deliver when implementing an algorithm, investigate the strategies organizations use to provide public values using an algorithm, and work towards a framework that defines the organizational capabilities required to use algorithms to create public value.’
What will be the benefit for society if you succeed in your aim?
As noted, algorithms can make governance more efficient and effective, however, the use of algorithms can lead to new decision-making errors. This is important, erroneous government decisions can have a major impact on people's lives, as has become painfully clear in the Tax Administration child care benefit scandal. By giving organizations the capabilities to develop, implement, and sustain the use of an algorithm, this research will enable organizations to take advantage of algorithms while giving them the tools necessary to detect and prevent adverse effects.