Epistemology and Ethics of Machine Learning


In July 2020, Dr. Konstantin Genin joined the Cluster of Excellence 'Machine Learning' at the University of Tübingen as head of the Independent Research Group 'Epistemology and Ethics of Machine Learning'.


Research Interests

It is widely acknowledged that questions of scientific methodology depend on ethical ones. If an experiment is unethical, it ought not to be performed. If an algorithm is unfair, it ought not to be implemented. From this perspective, ethics responds to methodological advances by rushing to install new guard-rails. But ethical questions also depend on methodological ones. Whether an experiment is ethical depends on whether similarly reliable inferences could be made from non-experimental data. Whether an algorithm is fair depends on how well it manages delicate tradeoffs between competing explications of fairness. The answers to these questions typically turn on methodological ones and -- more often than not -- these are both highly technical and hotly contested. From this perspective, methodological advances lead inevitably to ethical ones. The goal of the "Ethics and Epistemology" research group is to work these problems from both sides: to approach methodological issues with an eye to their social consequences and to approach ethical issues with an eye to methodological solutions.


Further Information

For further information see Konstantin Genin's Website.

The group has currently several open positions. For more information see our 'Open Positions' section.

About

Konstantin Genin has BAs in Mathematics and Philosophy from Brown University. He received a PhD in Logic, Computation and Methodology from the Philosophy Department at Carnegie Mellon University. His dissertation was supervised by Kevin Kelly. Before joining the cluster, he was a Postdoctoral Fellow in the Philosophy Department at the University of Toronto, supervised by Franz Huber.

Inspired by formal learning theory, Konstantin uses topological methods to investigate the inherent complexity of problems in statistics and machine learning. In particular, his work gives a non-circular epistemic justification for Ockham's razor in statistical inference. He is currently focused on applying these methods to problems in causal inference and algorithmic fairness.


Contact

Dr. Konstantin Genin
Epistemology and Ethics of Machine Learning

University Tübingen
Cluster of Excellence 'Machine Learning'
Maria-von-Linden-Str. 6, 4th floor
Room No. 40-5/A15
72076 Tübingen

+49 7071 2970892
konstantin.geninspam prevention@uni-tuebingen.de