Realising Value from AI-Enabled Decision Systems with Fair Outcomes:

An Explorary Case Study

Presented at FinanceCom2022, University of Twente.

Koefer, F., Lemken, I., & Pauls, J. (2023). Realising fair outcomes from Algorithm-Enabled Decision Systems: An Exploratory Case Study. In Van Hillegersberg, J., et al. (Eds.), Enterprise Applications, Markets and Services in the Finance Industry. Springer International Publishing.

An extended version of this paper has been published as a working paper with the European Investment Fund Research & Market Analysis working paper series under the title “Fairness in algorithmic decision systems: A microfinance perspective”

This project came to be as part of AI, Human Rights & Ethics, a joint course of the universities of Münster, Twente and Leiden. I collaborated with with Franziska Koefer (Twente) and Jan Pauls (Münster).

Fairness is a crucial concept in the context of artificial intelligence ethics and policy, and it is gaining more and more intention. However, it is difficult to implement high-level principles in practice. In our paper, we propose a newly developed frameworks intended to guide organizations to do this in a structured way, by using a lifecycle model. We take a holistic view by analyzing the challenges throughout a typical AI product life cycle while focusing on the critical question of how rather broadly defined fairness principles may be translated into day-to-day practical solutions at the organizational level. Our case study focuses on a microfinance organization that is using AI-enabled credit scoring to support the screening process to particularly financially marginalized entrepreneurs. This paper highlights the importance of considering the strategic role of the organization when developing and evaluating fair algorithm-enabled decision systems. Overall, the proposed framework and the results of this study can be used to inspire the right questions that suit the context an organization is situated in when implementing fair AI.

Igor

Life is full of waiting. Waiting for the water to boil, the bus to arrive, for a lecture to be over… However, unlike most machines, we are not truly good at waiting. This work, created together with Imani Dap and Ties Lind as part of the course Artificial Creatures, studies the quality of impatience in ourselves and in others. What is it that makes impatience different from nervousness or restlessness?

Insatiably Impatient (“Igor” for friends) is a simple robot with the body of a wind-up alarm clock and a ticking pen as an arm. It combines the rhythmic ticking of the clock, which reminds us of the passing of time, with the unpredictable semi-rhythmic ticking of a pen on the floor it’s standing on. This behavior, also known as fidgeting, makes Igor into an impatient creature.

Stories of a Real Man

During my Master's degree in Media Technology, I seized the opportunity to embark on an unconventional journey for my graduation project, capitalizing on the vast resources available to me within the program.

After insightful discussions with my supervisors, I chose to undertake my research via a hybrid autoethnography. The aim was to delve deeper into the formation of my understanding of maleness and masculinity. I did this through narrative writing and conducting qualitative interviews with individuals intimately connected to my life. To maintain the personal essence of my research, I deliberately chose to review the relevant literature after completing my own data collection and analysis.

I discovered various influential factors that shaped my perceptions, among which my individual personality, my upbringing, and my immersion in the orchestral subculture with its distinct, alternative norms. As the study focused on me personally, it helped me to gain a better understanding of myself and the topic of maleness and masculinity.

Hakuna Mutato

The Science to Experience Expo

MUTATE! The theme for the famous Media Technology science-to-experience exhibition in V2, Rotterdam was clearly inspired by the corona crisis, but our work went into a very different direction (like most MT works do).

Mark Magee, Ties Lind and I created an experience entitled Hakuna Mutato. Functioning as creators and hosts, we would start a conversation with the visitors of the exhibition. This conversation, that we could adjust to the knowledge of the visitor, was about physical base constants. These constants, despite their name, are redefined roughly once per decade. The focus of our work was on the second, because this unit is probably next to be redefined. However, instead of going from atomic to optical clocks, we decided to go by the visitor’s intuition. Every visitor got to play a game where they had to press a button every second. During this game, the colors on the screen and in the room were adjusted to be following their definition of the second. At the end, we would show how their second changed the world, where a liter of water could suddenly weigh 0.5 kg and a meter could suddenly be three times as long. As a little souvenir, the visitor got a string with the length of their “meter” showing how all these definitions are intertwined.

Limme Ricky

During one of my most compelling academic courses, we investigated the notion of computational creativity by conceptualizing and realizing three distinct projects, each incorporating both a pragmatic and theoretical component. These projects encompassed the creation of a cookie recipe generator (individual project), a poem generator, and a Co-Creative Composition Machine, the latter two being collaborative efforts with my colleague, Michael Grauwde.

One of these creations, dubbed Limme RICKY, is a pioneering limerick generator. Not only does it aim to produce limericks, but it also serves to critically interrogate and challenge the traditional form and conventions of limericks. This is accomplished through the use of Markov Chains, creating novel variations the structure of the limerick and pushing the boundaries of this poetic form.

Cici

During one of my most compelling academic courses, we investigated the notion of computational creativity by conceptualizing and realizing three distinct projects, each incorporating both a pragmatic and theoretical component. These projects encompassed the creation of a cookie recipe generator (individual project), a poem generator, and a Co-Creative Composition Machine, the latter two being collaborative efforts with my colleague, Michael Grauwde.

Cici is an innovative web platform that utilizes the MusicRNN model to enable users to collaboratively contribute to an evolving, infinite song. Representing a synthetic composer, Cici traverses the digital realm, inviting users to add new lines to the ongoing song. With user contributions saved on a central server, the project fosters a seamless, collective composition process. This offers a unique exploration of collaborative creativity, underpinned by machine learning algorithms.

Portfolio

If you want to see my complete portfolio, please get into contact with me