Shalom Chalson

The stakes in medicine are high. Crucial decisions, often involving matters of life and death, have to be made quickly. However, medicine is practised under great uncertainty. Clinicians sometimes lack the evidence required to diagnose or treat patients with confidence; patients struggle to know when to seek care; and policymakers must design rules or plan interventions using incomplete data. When faced with a decision problem, it is often unclear what the right course of action is or how to respond when something goes wrong. But given the stakes, delaying action is often not an option. Artificial intelligence (AI) is transforming our epistemic and moral landscapes. With access to vast datasets and the computational power required to analyse them, AI detects patterns that clinicians would miss, quantifies policy trade-offs that were previously invisible and generates increasingly fine-grained predictions. However, better evidence does not necessarily mean the resolution of moral uncertainty. With more information at our disposal, we may have to make more morally significant choices than ever before. We must also assign responsibility in contexts where decision-making is distributed across both humans and AI agents, even though the latter may at times operate in ways entirely opaque to us. This special e-issue of the Journal of Medical Ethics brings together nine articles that address AI’s transformation of uncertainty in medicine. This collection is unified by a central insight: AI may broaden our access to information and offer new, optimised solutions to longstanding problems, but its use introduces new complications to already challenging moral quandaries. Indeed, with AI in the loop, how to choose or act permissibly and how to distribute moral responsibility in a complex system have become more difficult to determine. The articles here take one of two approaches to these issues. The first is to show how AI expands …

On 16 April 2025, the Supreme Court of the UK ruled that when interpreting the UK’s Equality Act (2010)—the Act of the UK Parliament that details protections against unlawful discrimination—the terms ‘man’, ‘woman’ and ‘sex’ pick out ‘biological sex’ (defined in the Court’s ruling as ‘the sex of a person at birth’) and not also ‘certificated sex’ (defined in the Court’s ruling as ‘the sex attained by the acquisition of a Gender Recognition Certificate’). Some have argued that the Court’s decision represents a significant setback for trans people, while others have welcomed the perceived clarity the decision has provided. The decision has had, and will continue to have, wide impact, such as on the lawful operation of single-sex spaces and services in the UK. Scrutiny of the Court’s judgement is therefore warranted. The judgement was based on an interpretation of what Parliament intended in the Equality Act, but there are unresolved questions: (1) Is the Court’s interpretation of what Parliament intended the terms ‘man’, ‘woman’ and ‘sex’ to refer to correct? (2) Was the Court’s ruling made in a sufficiently just way? With these questions in mind, this article has two main aims. First, to summarise the Court’s recent decision and its social impacts. Second, to stress the importance of procedural justice in evaluating this verdict. This article thereby presents a preliminary case for the adequate consultation of all relevant stakeholders in potentially contentious legal decisions—especially where significant practical consequences for different social groups are likely.

This thesis is on the topic of discrimination. In the sense relevant to moral inquiry, discrimination refers to a wrongful sort of differential treatment. Agents discriminate in treating some differently to others, and in ways that require restitution. My main aim is to develop a moralised conception of discrimination with an eye on rational decision-making. On the view presented in this thesis, an agent discriminates in engaging in differential treatment on the basis of considerations that are not sufficiently relevant to their legitimate goals. Agents can do so by misconstruing their evidence or by adopting goals that prescribe wrongful differential treatment. At first glance, this view runs counter to intuition. A greater focus on rationality dilutes discrimination's putative moral content. I show that the view presented allows that some cases of discrimination—as marked by socially salient attributes, for instance—warrant particular moral concern or are especially wrong. In the course of this work, I pursue two subsidiary aims. One, to analyse discrimination in a way that is neutral between moral theories. And two, to develop an account of discrimination that extends to a broad range of inequalities, including ones not precipitated by background injustice. The upshot is a theory of discrimination capable of guiding moral inquiry under a range of variable contextual factors.

We argue in favor of affirmative action. There are two central points to our argument. First, if two or more candidates for a position are matched in competence, then one ought to prefer a candidate from a disadvantaged, disenfranchised, or minority background rather than defer to the outcome of a lottery. Second, a closer look at actual selection procedures reveals that situations where candidates are matched in competence are much more common than one might think. Our argument results in a robust justification for affirmative action from minimally controversial premises.

Generative artificial intelligence systems based on transformers, including both text-generators like GPT-4 and image generators like DALL-E 3, have recently entered the popular consciousness. These tools, while impressive, are liable to reproduce, exacerbate, and reinforce extant human social biases, such as gender and racial biases. In this paper, we systematically review the extent to which DALL-E Mini suffers from this problem. In line with the Model Card published alongside DALL-E Mini by its creators, we find that the images it produces tend to represent dozens of different occupations as populated either solely by men (e.g., pilot, builder, plumber) or solely by women (e.g., hairdresser, receptionist, dietitian). In addition, the images DALL-E Mini produces tend to represent most occupations as populated primarily or solely by White people (e.g., farmer, painter, prison officer, software engineer) and very few by non-White people (e.g., pastor, rapper). These findings suggest that exciting new AI technologies should be critically scrutinized and perhaps regulated before they are unleashed on society.