Fricker shows that virtue epistemology provides a general epistemological idiom in which these issues can be forcefully discussed.

In the last several months, several major disciplines have started their initial reckoning with what ChatGPT and other Large Language Models (LLMs) mean for them – law, medicine, business among other professions. With a heavy dose of humility, given how fast the technology is moving and how uncertain its social implications are, this article attempts to give some early tentative thoughts on what ChatGPT might mean for bioethics. I will first argue that many bioethics issues raised by ChatGPT are similar (...) to those raised by current medical AI – built into devices, decision support tools, data analytics, etc. These include issues of data ownership, consent for data use, data representativeness and bias, and privacy. I describe how these familiar issues appear somewhat differently in the ChatGPT context, but much of the existing bioethical thinking on these issues provides a strong starting point. There are, however, a few “new-ish” issues I highlight – by new-ish I mean issues that while perhaps not truly new seem much more important for it than other forms of medical AI. These include issues about informed consent and the right to know we are dealing with an AI, the problem of medical deepfakes, the risk of oligopoly and inequitable access related to foundational models, environmental effects, and on the positive side opportunities for the democratization of knowledge and empowering patients. I also discuss how races towards dominance (between large companies and between the U.S. and geopolitical rivals like China) risk sidelining ethics. (shrink)

This theoretical paper considers the morality of machine learning algorithms and systems in the light of the biases that ground their correctness. It begins by presenting biases not as a priori negative entities but as contingent external referents—often gathered in benchmarked repositories called ground-truth datasets—that define what needs to be learned and allow for performance measures. I then argue that ground-truth datasets and their concomitant practices—that fundamentally involve establishing biases to enable learning procedures—can be described by their respective morality, here (...) defined as the more or less accounted experience of hesitation when faced with what pragmatist philosopher William James called “genuine options”—that is, choices to be made in the heat of the moment that engage different possible futures. I then stress three constitutive dimensions of this pragmatist morality, as far as ground-truthing practices are concerned: the definition of the problem to be solved, the identification of the data to be collected and set up, and the qualification of the targets to be learned. I finally suggest that this three-dimensional conceptual space can be used to map machine learning algorithmic projects in terms of the morality of their respective and constitutive ground-truthing practices. Such techno-moral graphs may, in turn, serve as equipment for greater governance of machine learning algorithms and systems. (shrink)

Artificial Intelligence in the form of different machine learning models is applied to Big Data as a way to turn data into valuable knowledge. The rhetoric is that ensuing predictions work well—with a high degree of autonomy and automation. We argue that we need to analyze the process of applying machine learning in depth and highlight at what point human knowledge production takes place in seemingly autonomous work. This article reintroduces classification theory as an important framework for understanding such seemingly (...) invisible knowledge production in the machine learning development and design processes. We suggest a framework for studying such classification closely tied to different steps in the work process and exemplify the framework on two experiments with machine learning applied to Facebook data from one of our labs. By doing so we demonstrate ways in which classification and potential discrimination take place in even seemingly unsupervised and autonomous models. Moving away from concepts of non-supervision and autonomy enable us to understand the underlying classificatory dispositifs in the work process and that this form of analysis constitutes a first step towards governance of artificial intelligence. (shrink)

Ethnographic study of the constitution of algorithms.