Data, Algorithmic Systems and Ethics

Algorithmic systems have immense possibilities in various fields but can have profound impacts on users and societies. Calls for and efforts towards the ethical design, development, deployment and regulation of algorithmic systems are growing along with demands to make them beneficial to "the common good". However, in a world defined by asymmetric power relations, it is important to ask: who decides what is good and ethical, and conversely, who gets the "good" done by someone else.

The research group “Data, Algorithmic Systems and Ethics” investigates social as well as ethical-technical questions around the just, fair, and transparent production and application of training data and algorithmic systems, with special focus on social, labor, and institutional contexts that shape them.

The central questions are:

• How can data-driven algorithmic systems be made just, fair, transparent, and beneficial to the common good on an ethical-technical basis? What does this mean for participation, inclusion and sharing?
• How do specific industrial and institutional contexts and working conditions influence the production of training data and algorithms? What are the implications for technology-related inequities?

We understand algorithmic systems as part of larger socio-technical systems and look for principles, architectures, and parameters that can lead to desired goals, and methods for determining those goals themselves. Especially relevant to our research focus are socio-technical systems based on Big Data, machine learning, or algorithm-based prediction and decision processes. To this end, we combine and apply research methods from computer science (e.g. modeling, analysis and implementation) and social sciences (e.g. ethnographic fieldwork and interviewing) as well as participatory design methodologies.

In the research group, we consider the tension between the challenges that seem inherent to data-driven systems (and thus, implicitly, equal across people and/or contexts) and the differential challenges that can carry, reinforce, or cause injustice. Truly common-good AI must consider both phenomena to make socio-technical systems "inclusive" for diverse groups and individuals – especially those “on the margins” – and aspire to enable participation by all. We are currently completing the setup of the research group and working on the concretization of individual projects. More details on concrete projects within the research group will follow.