•  71
    Many AI systems that make important decisions are black boxes: how they function is opaque even to their developers. This is due to their high complexity and to the fact that they are trained rather than programmed. Efforts to alleviate the opacity of black box systems are typically discussed in terms of transparency, interpretability, and explainability. However, there is little agreement about what these key concepts mean, which makes it difficult to adjudicate the success or promise of opacit…Read more
  •  160
    Pursuit and inquisitive reasons
    Studies in History and Philosophy of Science Part A 94 17-30. 2022.
    Sometimes inquirers may rationally pursue a theory even when the available evidence does not favor that theory over others. Features of a theory that favor pursuing it are known as considerations of promise or pursuitworthiness. Examples of such reasons include that a theory is testable, that it has a useful associated analogy, and that it suggests new research and experiments. These reasons need not be evidence in favor of the theory. This raises the question: what kinds of reasons are provided…Read more
  •  267
    Responsibility for Collective Epistemic Harms
    Philosophy of Science 1-41. forthcoming.
    Discussion of epistemic responsibility typically focuses on belief formation and actions leading to it. Similarly, accounts of collective epistemic responsibility have addressed the issue of collective belief formation and associated actions. However, there has been little discussion of collective responsibility for preventing epistemic harms, particularly those preventable only by the collective action of an unorganized group. We propose an account of collective epistemic responsibility which …Read more
  •  40
    RAWLSNET: Altering Bayesian Networks to Encode Rawlsian Fair Equality of Opportunity
    with David Liu, Zohair Shafi, Tina Eliassi-Rad, and Scott Alfeld
    Proceedings of the 2021 AAAI/ACM Conference on AI, Ethics, and Society. 2021.
    We present RAWLSNET, a system for altering Bayesian Network (BN) models to satisfy the Rawlsian principle of fair equality of opportunity (FEO). RAWLSNET's BN models generate aspirational data distributions: data generated to reflect an ideally fair, FEO-satisfying society. FEO states that everyone with the same talent and willingness to use it should have the same chance of achieving advantageous social positions (e.g., employment), regardless of their background circumstances (e.g., socioecono…Read more
  •  297
    What's Fair about Individual Fairness?
    Proceedings of the 2021 AAAI/ACM Conference on AI, Ethics, and Society. 2021.
    One of the main lines of research in algorithmic fairness involves individual fairness (IF) methods. Individual fairness is motivated by an intuitive principle, similar treatment, which requires that similar individuals be treated similarly. IF offers a precise account of this principle using distance metrics to evaluate the similarity of individuals. Proponents of individual fairness have argued that it gives the correct definition of algorithmic fairness, and that it should therefore be prefer…Read more
  •  245
    Fragmentation and Old Evidence
    Episteme 1-26. forthcoming.
    Bayesian confirmation theory is our best formal framework for describing inductive reasoning. The problem of old evidence is a particularly difficult one for confirmation theory, because it suggests that this framework fails to account for central and important cases of inductive reasoning and scientific inference. I show that we can appeal to the fragmentation of doxastic states to solve this problem for confirmation theory. This fragmentation solution is independently well-motivated because of…Read more
  •  485
    Publishing without (some) belief
    Thought: A Journal of Philosophy 9 (4): 237-246. 2020.
    Thought: A Journal of Philosophy, EarlyView.
  •  319
    How to endorse conciliationism
    Synthese 198 (10): 9913-9939. 2021.
    I argue that recognizing a distinct doxastic attitude called endorsement, along with the epistemic norms governing it, solves the self-undermining problem for conciliationism about disagreement. I provide a novel account of how the self-undermining problem works by pointing out the auxiliary assumptions the objection relies on. These assumptions include commitment to certain epistemic principles linking belief in a theory to following prescriptions of that theory. I then argue that we have indep…Read more
  •  543
    Endorsement and assertion
    Noûs 55 (2): 363-384. 2021.
    Scientists, philosophers, and other researchers commonly assert their theories. This is surprising, as there are good reasons for skepticism about theories in cutting-edge research. I propose a new account of assertion in research contexts that vindicates these assertions. This account appeals to a distinct propositional attitude called endorsement, which is the rational attitude of committed advocacy researchers have to their theories. The account also appeals to a theory of conversational pra…Read more
  •  290
    Method Coherence and Epistemic Circularity
    Erkenntnis 84 (2): 455-480. 2019.
    Reliabilism is an intuitive and attractive view about epistemic justification. However, it has many well-known problems. I offer a novel condition on reliabilist theories of justification. This method coherence condition requires that a method be appropriately tested by appeal to a subject’s other belief-forming methods. Adding this condition to reliabilism provides a solution to epistemic circularity worries, including the bootstrapping problem.
  •  478
    Rational endorsement
    Philosophical Studies 175 (10): 2649-2675. 2018.
    It is valuable for inquiry to have researchers who are committed advocates of their own theories. However, in light of pervasive disagreement, such a commitment is not well explained by the idea that researchers believe their theories. Instead, this commitment, the rational attitude to take toward one’s favored theory during the course of inquiry, is what I call endorsement. Endorsement is a doxastic attitude, but one which is governed by a different type of epistemic rationality. This inclusive…Read more
  •  247
    Virtue epistemology has been divided into two camps: reliabilists and responsibilists. This division has been attributed in part to a focus on different types of virtues, viz., faculty virtues and character virtues. I will argue that this distinction is unhelpful, and that we should carve up the theoretical terrain differently. Making several better distinctions among virtues will show us two important things. First, that responsibilists and reliabilists are actually engaged in different, comple…Read more