Contested futures of AI ethics for global population health

Abstract

Ethical debate about artificial intelligence (AI) for health is fundamental to the good governance and long-term success of this group of technologies in healthcare and public health. Building on foundational contributions on this topic from the field of bioethics, we propose that interdisciplinary work on AI ethics should deepen the already strong engagements that exist with Science and Technology Studies (STS) to uphold core commitments of global population health. STS is an interdisciplinary field that specifically explores the meaning, development and governance of science and technology and offers enormous conceptual resources for AI ethics. We present a path towards deeper integration of STS and bioethics for an AI ethics that can more meaningfully address the most important issues presented by AI for global population health.

WHAT IS ALREADY KNOWN ON THIS TOPIC.

WHAT THIS STUDY ADDS

• This study describes an approach to AI ethics that is informed by insights from Science and Technology Studies (STS) to more adequately address the pillars of Universal Health Coverage, planetary health and local values.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• This study can inform emerging directions in AI ethics for health that promote deeper engagement from AI governance with foundational issues illuminated by the framework bringing bioethics and STS together.

The global contexts of artificial intelligence (AI) in healthcare and public health have shifted in important ways over the past 10 years. Evidence is mounting of the positive impacts of narrow implementations of AI in specific health-related contexts,^1,2 along with insights related to the hidden harms that might accompany their deployment.^3,4 The hype of earlier AI technologies has given way to a newer wave of hype over generative AI and the prospects of further advances in multimodal models and agentic AI for impacts on health and healthcare.⁵ The global geopolitics of AI have also shifted, representing divergent views on how much regulation is warranted over innovation in the interest of public protection.^6,7 The deregulatory shift in the USA co-exists with Chinese leadership in publications and patents on AI,⁸ while the 2025 AI Action Summit in Paris introduced deeply divergent approaches to AI governance. Concentrated power in large technology corporations (‘Big Tech’) enables their immense influence on policy internationally.⁹ Along with these shifts, thought and practice of AI ethics must also evolve, seeking out comprehensive frameworks capable of identifying and mitigating against systemic long-term risks of AI for population health.

In this article, we argue that an interdisciplinary approach to AI ethics at the intersection of bioethics and Science and Technology Studies (STS) presents the most promising path towards an AI ethics for health capable of protecting population health for future generations. Drawing on past reviews and arguments integrating various positions about AI from the field of bioethics,^10–13 we propose a path for AI ethics that can confront these challenges in explicit and theoretically sophisticated ways. Bioethics is essential for the guidance it provides in navigating between the conflicting normative commitments that shape dialogue on topics such as AI, health and good governance.¹⁴ However, in a moment where those with political authority wield immense power over what counts as ‘truth’,¹⁵ the field must also grapple with enormous tensions over what is at stake in debates on AI ethics and health.

STS is perhaps the field in which these tensions have been most central, offering crucial resources for exploring the sources of normative commitments alongside debates about facts and values. Building on the effort in STS to move beyond counterproductive dualistic tensions (eg, between ‘right’ and ‘wrong’ notions in AI ethics),^11,16,17 we present an approach based on three practical commitments that characterise global efforts to promote population health: (1) respect for local cultures and norms, (2) commitment to Universal Health Coverage (UHC) and (3) efforts to promote planetary health. We suggest that such an approach synthesises what is most valuable from both bioethics and STS for an ethics of AI that is more explicitly committed to securing global population health for future generations.

Bioethics meets STS

Bioethicists have made foundational contributions to the evolving field of AI ethics in healthcare and public health. Bioethicists have emphasised the importance of considering ethical issues along the lifecycle of AI technologies, from conceptualisation through to evaluation, sustainability or decommissioning.¹⁸ Contributions from bioethics have also illustrated the many levels of analysis at which ethical issues demand attention, from the individual to the societal.¹⁹ Furthermore, bioethics has supported a detailed assessment of strategies to align AI with ethical values in their development, deployment and governance.²⁰ This latter observation about bioethics, its commitment to informing decisions about ethically contentious phenomena, makes it indispensable in the ongoing dialogue on AI for health and healthcare. Bioethics is often imminently practical for clinical and policy decision-making.

Despite its practicality, bioethics as applied to AI has also been critiqued. For example, commentators have suggested that applied work in bioethics tends to assume a universal common morality, neglecting the nuances of cultures and norms that shape local views about ethics.¹¹ Specifically in relation to AI and other health technologies, bioethics has also been critiqued for accepting the boundaries drawn around ethical dialogue on new technologies as established by proponents of those technologies.¹¹ Such boundaries tend to steer dialogue away from deeper criticisms. These critiques are certainly not true of all approaches to bioethics, but they help to illustrate the value of drawing on a field that is focused specifically on the complex normative, epistemic and ontological dimensions of science and technology: STS.

STS is a rich and diverse interdisciplinary field, bringing together scholars from engineering, computer science, biology, history, philosophy, sociology, anthropology, political science, economics and others to study science and technology. STS examines the institutions and practices that create and maintain technologies, alongside the meaning and consequences of those technologies in the world.²¹ A primary guiding question in STS revolves around how technologies are made, referring not only to design decisions but also to the ways that technologies are connected to controversies beyond the material artefact itself. Drawing on emerging currents in interdisciplinary theory, STS rejects conventional boundaries around science and technology, instead pursuing the networks of people and things that give these phenomena their meaning.²² In this way, STS subjects the constitution of technologies such as AI to close empirical scrutiny, holding in question its assumed capacities and potential value. In the same way, STS also subjects the practice of bioethics to close empirical scrutiny, questioning the way it conceptualises its objects of focus and the analytic techniques upon which it relies to arrive at practical normative recommendations.

Although STS generally does not take a strongly normative orientation, the relationship between bioethics and STS is not new. Researchers in STS have explored how technologies produce new demands on ethical decision-making,²³ how concepts such as ‘care’ shape our ethical obligations in science²⁴ and how values can be built into design processes.²⁵ Rather than producing normative argumentation for the centrality of a particular set of ethical commitments, this work tends to observe practices in a field and explore the impacts of those practices on the concepts, scientific results and technologies produced. Doing so generates empirically grounded insights that open up new spaces to imagine the evolving ethical dimensions of a given domain of life and explore new ethical challenges and possibilities. It is through the commitment to empirical grounding and a diverse body of theory that is closely tied to science and technology that STS enables bioethics to examine new ethically salient dimensions of AI for health. The role for bioethics in this context is to examine competing assumptions and value systems illuminated through empirical work, articulate an approach to adjudicating between them that is informed by theory and support the process of obtaining an ethical judgement to guide action.

The approach we describe here exists in relation to a history of applied ethics work on health-related innovations, for example, through the development of interdisciplinary work on the ethical, legal and social implications (or aspects) of scientific and technological developments (often referred to as ‘ELSI’ or ‘ELSA’).²⁶ ELSI work remains an active domain of scholarship, in part through the emergence of responsible innovation as an orientation to the governance of emerging technologies. However, we suggest that the field requires an approach that integrates considerations such as ethical, legal and social issues into a synthesised understanding of the practices through which technologies are developed and governance decisions are made. We contend that these are not simply different domains to be analysed with different methods but rather must be understood together as part of a common world to inform meaningful AI ethics for health.²⁷

Drawing on these insights, we position three longstanding commitments from global work on population health as foundational to our argument. First, moving away from a clear sense of a universal morality, here we highlight the central importance of the local cultures and norms that produce locally driven ethics. Second, drawing on the longstanding commitment to establish systems that ensure all members of a population have access to healthcare, we position UHC as central to the future of AI ethics for health. Third, inspired by efforts in STS to expand understandings of the many actors and contextual influences that shape science and technology, we build planetary health into this approach as a final pillar. Planetary health demands close attention to the concentration of power and decision-making in the hands of a small group of governments and corporations whose interests are served by investments in the AI industry more generally.²⁸ Integrating these three foundational commitments, we present two fundamental concepts from STS and the questions they would pose in efforts to understand and promote the ethics of AI for the global population’s health. Please see the figure 1 for a depiction of how these three features interact with AI ethics for global public health.

Figure 1. Illustrates how the ideas we have introduced fit together, posing a series of questions in relation to each idea that can guide ethical dialogue about AI for global population health. These questions are intended to encourage deeper reflection on the state of AI in the world, informing ethical debate about its deployment and governance on a global scale. AI, artificial intelligence.

Infrastructures and imaginaries

To illustrate the value of STS to the ethics of AI for health, we suggest that two major topics of study in STS are especially helpful in addressing the evolving ethical issues in AI for health: ‘infrastructures’ and ‘imaginaries’. In STS, infrastructures have garnered a great deal of attention as often ignored but foundational features of everyday life.²⁹ Rather than seeing infrastructures as static structures in the background, STS treats infrastructures as ‘relational’—as being constituted through sets of relational practices over time.²⁹ In relation to health AI, infrastructures are composed of material elements such as computational resources, data storage systems and energy consumption, as well as human labour, including that of patients, software developers and healthcare professionals. For example, classification systems are a kind of infrastructure (eg, diagnostic categories used in some algorithmic decision-making models) that play a fundamental role in structuring how AI operates within healthcare settings.

In the contemporary world, people interact with interlocking standards, conventions and technologies, all of which are built into infrastructures that shape everyday life.³⁰ Infrastructures are connected to one another, producing a network of elements such as information repositories of electronic health records, diagnostic criteria and workflows in healthcare organisations. This network of standards, technologies and conventions codifies and prescribes ethics and values, which are consequential for the work of healthcare and public health. Examining these infrastructures enables AI ethics to critically assess not only their technical efficiency but also their environmental, social and political implications, which we suggest are essential for meaningful AI ethics for the future of population health.

In STS, the concept of imaginaries refers to collectively envisioned forms of life and social order that become materialised in the design and implementation of specific policies, human habits and social expectations.³¹ Early studies of imaginaries explored how government assumptions about science and technology came to shape policy and investment in particular ways over time.³¹ The study of imaginaries is based on the premise that narratives about the world and how we ought to live in the world have a major impact on technological innovation and development. This line of thought is important with respect to AI for health, given the enormous hype and financial investment that characterises this class of technologies. The concept of imaginaries in applied ethics makes it possible to move beyond a dominant analysis of the value of AI for health based solely on costs and benefits. Instead, invoking imaginaries allows for the ethical consideration of the expectations, hopes and fears that characterise how we understand ‘good health’ and its relationship to AI; imaginaries encourage us to ask the question, ‘does AI help to produce the world we want to live in?’ In so doing, it draws deeper ethical attention to the assumptions we make about what AI can do for health as we debate its governance for healthcare and population health.

Conclusion

AI is a central figure in the intersecting democratic, economic and governance crises that interfere with efforts to secure global population health for future generations. We suggest that more radical thought and practice of AI ethics is necessary to adequately critique AI and its relationship to public health and healthcare. Only through deeper critical reflection, not only on the technology but also on the network of relations in which it is embedded, can its positive role for population health be more clearly advanced. We suggest that an approach bringing bioethics and STS together is the most promising path to achieve this goal.