Algorithmic predictions are promising for insurance companies to develop personalized risk models for determining premiums. In this context, issues of fairness, discrimination, and social injustice might arise: Algorithms for estimating the risk based on personal data may be biased towards specific social groups, leading to systematic disadvantages for those groups. Personalized premiums may thus lead to discrimination and social injustice. It is well known from many application fields that such biases occur frequently and naturally when prediction models are applied to (...) people unless special efforts are made to avoid them. Insurance is no exception. In this paper, we provide a thorough analysis of algorithmic fairness in the case of insurance premiums. We ask what “fairness” might mean in this context and how the fairness of a premium system can be measured. For this, we apply the established fairness frameworks of the fair machine learning literature to the case of insurance premiums and show which of the existing fairness criteria can be applied to assess the fairness of insurance premiums. We argue that two of the often-discussed group fairness criteria, independence (also called statistical parity or demographic parity) and separation (also known as equalized odds), are not normatively appropriate for insurance premiums. Instead, we propose the sufficiency criterion (also known as well-calibration) as a morally defensible alternative that allows us to test for systematic biases in premiums towards certain groups based on the risk they bring to the pool. In addition, we clarify the connection between group fairness and different degrees of personalization. Our findings enable insurers to assess the fairness properties of their risk models, helping them avoid reputation damage resulting from potentially unfair and discriminatory premium systems. (shrink)

There is a very plausible principle of the transitivity of justifying reasons. It says that if "p" is better justified than "q" (all things considered) and "q" better than "r", then "p" is better justified than "r" (all things considered). There is a corresponding principle of rational theory choice. Call one theory "a better theory than" another theory if all criteria of theory choice considered (explanatory power, simplicity, empirical adequacy, etc.), the first theory meets the criteria better than the second (...) theory. The corresponding transitivity principle says that if theory A is a better theory than theory B and if theory B is a better theory than theory C, then A is a better theory than theory C. I argue against this principle. It turns out that whenever there are 2 or more relevant and independent criteria of theory evaluation, and whenever at least of one the criteria is "non-linear" in a certain sense, there may be violations of transitivity which do not violate any standards of rationality (of theory choice). This shows, again, that theory choice cannot be seen as merely the application of given rules of rational theory choice. (shrink)

Introduction to and overview over my book "Epistemic Contextualism. A Defense" (OUP 2016).

Ricoeur has written about translation as an ethical paradigm. Translation from one language to another, and within one’s own language, provides both a metaphor and a real mechanism for explaining oneself to the other. Attempting and failing to achieve symmetry between two languages is a manifestation of the asymmetry inherent in human relationships. If actively pursued, translation can show us how to forgive other people for being different from us and thus serves as a paradigm for tolerance. In full acceptance (...) that this will be impossible, Ricoeur uses the model of translation as a way of understanding European integration, with three aspects: translation, shared narrative and shared forgiveness of Europe’s history. These models provide a strong statement about tolerance and become even more significant through their conversation with the negativity that suffuses them. He draws on his knowledge of psychoanalysis to explain that the translator suffers through remembering and through mourning the loss of perfection; there must be acknowledgement of deficiency. This acceptance of imperfection and of limits to success is a key element in Ricoeur’s philosophy and is explored from the 1950s onwards in his study of negativity; denied by phenomenology and explored by Hegel. Negation is vital for understanding the world, but it can preclude us from access to meaning when it becomes negativity. Translation can provide the bridge to span the tension between the pathology of denial and different interpretations, and projection of evil into others, which I believe is at the heart of the perceived incompatibilities between Islam and the West. There is a political urgency to this enterprise, given the ‘othering’ of the Muslim world that has replaced the Cold War dichotomies between Communist as ‘other’ and the capitalist world. References to the Muslim as the current ‘other’ will be part of my discussion. As well as seeking to understand Ricoeur’s model of translation, we will examine whether his model works in a world where many speak no Arabic, Urdu or Farsi, or indeed whether it has any relevance for people who do not translate from one language to another. (shrink)

Review of Fodor, "Hume Variations".

In response to the COVID-19 pandemic, countries around the world developed ICU triaging guidelines. One important point of convergence was an explicit commitment to...