Robert Shanklin

An Artificial Intelligence algorithm trained on data that reflect racial biases may yield racially biased outputs, even if the algorithm on its own is unbiased. For example, algorithms used to schedule medical appointments in the USA predict that Black patients are at a higher risk of no-show than non-Black patients, though technically accurate given existing data that prediction results in Black patients being overwhelmingly scheduled in appointment slots that cause longer wait times than non-Black patients. This perpetuates racial inequity, in this case lesser access to medical care. This gives rise to one type of Accuracy-Fairness trade-off: preserve the efficiency offered by using AI to schedule appointments or discard that efficiency in order to avoid perpetuating ethno-racial disparities. Similar trade-offs arise in a range of AI applications including others in medicine, as well as in education, judicial systems, and public security, among others. This article presents a framework for addressing such trade-offs where Machine Learning and Optimization components of the algorithm are decoupled. Applied to medical appointment scheduling, our framework articulates four approaches intervening in different ways on different components of the algorithm. Each yields specific results, in one case preserving accuracy comparable to the current state-of-the-art while eliminating the disparity.

We often go out to the movies, theater, or ballet, preferring to share the experience with others rather than watch at home alone. We do the same with food and drink, for instance, by going to tasting rooms to sample wines and talk with others. To have these sorts of experiences, we plan, coordinate, and engage in a range of complex social practices. These practices often lead to the formation of audiences, and philosophers since Aristotle have argued that the success of certain art forms depends on audience reactions.1 In addition, many processes of aesthetic education involve individuals learning about aesthetic objects and judgments together. If so, then understanding how groups experience aesthetic objects...

The internalist-externalist debate about semantic and mental contents concerns whether the contents of certain claims and beliefs depend on facts external to the people having those beliefs or not. However, rather than just join up with either side, I argue for re-casting the debate so as to allow for hybrid internalist-externalist views, on the grounds that such views can help explain certain phenomena associated with slurs and pejoratives. If the debate can indeed be recast in this way and if hybrid views offer significant explanatory power, then they deserve further exploration.

This paper concerns the semantic difference between strong and weak neces-sity modals. First we identify a number of explananda: their well-known in-tuitive difference in strength between ‘must’ and ‘ought’ as well as differ-ences in connections to probabilistic considerations and acts of requiring and recommending. Here we argue that important extant analyses of the se-mantic differences, though tailored to account for some of these aspects, fail to account for all. We proceed to suggest that the difference between ’ought’ and ’must’ lies in how they relate to scalar and binary standards. Briefly put, must(φ) says that among the relevant alternatives, φ is selected by the relevant binary standard, whereas ought(φ) says that among the relevant al-ternatives, φ is selected by the relevant scale. Given independently plausi-ble assumptions about how standards are provided by context, this ex-plains the relevant differences discussed.