About
I am an Assistant Professor in the Department of Philosophy at the University of Utah. I approach methodological issues in the data sciences with an eye to their social consequences. Fundamentally, I am interested in the relationship between methodological and social progress. Formerly, I led the Research Group: Epistemology and Ethics of Machine Learning at the University of Tübingen. Before that, I was a postdoctoral fellow in Philosophy at the University of Toronto supervised by Franz Huber. My dissertation, supervised by Kevin T. Kelly in Philosophy at Carnegie Mellon, is available here.
This is my CV and google scholar. Contact me at konstantin [dot] genin [at] gmail.
Research
philosophy of science, statistics and machine learning, inductive inference, algorithmic fairness, Ockham's razor, methodology, formal epistemology, learning theory, causal discovery, topology
Many of my projects involve predicting and shaping the social effects of AI systems. That includes contributing to the design of randomized trials in medical AI (Genin and Grote, 2021); studying the distributive effects of algorithmic public administration (Zezulka and Genin, 2024); examining the evolving notion reliability in machine learning (Grote, Genin and Sullivan, 2024); and analyzing how computational psychiatry is reshaping our understanding of mental disorder (Genin, Grote and Wolfers, 2024). My current projects include (i) exploring how social-statistical predictions might better support deliberation in policy contexts; (ii) evaluating the effect of algorithmic prediction of gender-based violence (iii) evaluating whether advances in causal discovery warrant a rethinking of the ethical trade-offs traditionally associated with randomized trials and (iv) interrogating the methodological foundations of polygenic risk scores and implications for social science and policy.
My work is founded on a learning-theoretic perspective on statistical discovery (Genin, 2022; Genin and Kelly, 2018) and formal epistemology (Genin and Huber, 2020). In my dissertation, I provide a topological complexity theory for statistical inference and a non-circular epistemic justification for Ockham's razor. These results are used to analyze the inherent difficulty of causal discovery (Genin, 2021; Genin and Mayo-Wilson, 2022). In collaboration with statisticians and computer scientists, the results of my dissertation have been generalized and applied to conditional independence testing (Boeken, Skapinakis, Genin and Mooij, 2026).