ICTworks

WHO recognizes that Artificial Intelligence (AI) holds great promise for the practice of public health and medicine. WHO also recognizes that, to fully reap the benefits of AI, ethical challenges for health care systems, practitioners and beneficiaries of medical and public health services must be addressed. Many of the ethical concerns described in this report predate the advent of AI, although AI itself presents a number of novel concerns.

Whether AI can advance the interests of patients and communities depends on a collective effort to design and implement ethically defensible laws and policies and ethically designed AI technologies. There are also potential serious negative consequences if ethical principles and human rights obligations are not prioritized by those who fund, design, regulate or use AI technologies for health.

AI’s opportunities and challenges are thus inextricably linked.

AI can enable resource-poor countries, where patients often have restricted access to health-care workers or medical professionals, to bridge gaps in access to health services. AI systems must be carefully designed to reflect the diversity of socio-economic and health-care settings and be accompanied by training in digital skills, community engagement and awareness-raising.

Systems based primarily on data of individuals in high-income countries may not perform well for individuals in low- and middle-income settings. Country investments in AI and the supporting infrastructure should therefore help to build effective health-care systems by avoiding AI that encodes biases that are detrimental to equitable provision of and access to health-care services.

The WHO Guidance Document, “Ethics and Governance of Artificial Intelligence for Health” was produced jointly by WHO’s Health Ethics and Governance unit in the department of Research for Health and by the department of Digital Health and Innovation. It is based on the collective views of a WHO Expert Group on Ethics and Governance of AI for Health, which comprised 20 experts in public health, medicine, law, human rights, technology and ethics.

The group analysed many opportunities and challenges of AI and recommended policies, principles and practices for ethical use of AI for health and means to avoid its misuse to undermine human rights and legal obligations.

AI for health has been affected by the COVID-19 pandemic. Although the pandemic is not a focus of this report, it has illustrated the opportunities and challenges associated with AI for health. Numerous new applications have emerged for responding to the pandemic, while other applications have been found to be ineffective.

Several applications have raised ethical concerns in relation to surveillance, infringement on the rights of privacy and autonomy, health and social inequity and the conditions necessary for trust and legitimate uses of data-intensive applications. During their deliberations on this report, members of the expert group prepared interim WHO guidance for the use of proximity tracking applications for COVID-19 contact-tracing.

6 Key Ethical Principles: AI for Health

This report endorses a set of key ethical principles. WHO hopes that these principles will be used as a basis for governments, technology developers, companies, civil society and inter-governmental organizations to adopt ethical approaches to appropriate use of AI for health.

1. Protecting human autonomy

Use of AI can lead to situations in which decision-making power could be transferred to machines. The principle of autonomy requires that the use of AI or other computational systems does not undermine human autonomy. In the context of health care, this means that humans should remain in control of health-care systems and medical decisions.

Respect for human autonomy also entails related duties to ensure that providers have the information necessary to make safe, effective use of AI systems and that people understand the role that such systems play in their care. It also requires protection of privacy and confidentiality and obtaining valid informed consent through appropriate legal frameworks for data protection.

2. Promoting human well-being, safety, and public interest.

AI technologies should not harm people. The designers of AI technologies should satisfy regulatory requirements for safety, accuracy and efficacy for well-defined use cases or indications.

Measures of quality control in practice and quality improvement in the use of AI over time should be available. Preventing harm requires that AI not result in mental or physical harm that could be avoided by use of an alternative practice or approach.

3. Ensuring transparency, explainability and intelligibility.

AI technologies should be intelligible or understandable to developers, medical professionals, patients, users and regulators. Two broad approaches to intelligibility are to improve the transparency of AI technology and to make AI technology explainable.

Transparency requires that sufficient information be published or documented before the design or deployment of an AI technology and that such information facilitate meaningful public consultation and debate on how the technology is designed and how it should or should not be used. AI technologies should be explainable according to the capacity of those to whom they are explained.

4. Fostering responsibility and accountability.

Humans require clear, transparent specification of the tasks that systems can perform and the conditions under which they can achieve the desired performance. Although AI technologies perform specific tasks, it is the responsibility of stakeholders to ensure that they can perform those tasks and that AI is used under appropriate conditions and by appropriately trained people.

Responsibility can be assured by application of “human warranty”, which implies evaluation by patients and clinicians in the development and deployment of AI technologies. Human warranty requires application of regulatory principles upstream and downstream of the algorithm by establishing points of human supervision.

If something goes wrong with an AI technology, there should be accountability. Appropriate mechanisms should be available for questioning and for redress for individuals and groups that are adversely affected by decisions based on algorithms.

5. Ensuring inclusiveness and equity.

Inclusiveness requires that AI for health be designed to encourage the widest possible appropriate, equitable use and access, irrespective of age, sex, gender, income, race, ethnicity, sexual orientation, ability or other characteristics protected under human rights codes.

AI technology, like any other technology, should be shared as widely as possible. AI technologies should be available for use not only in contexts and for needs in high-income settings but also in the contexts and for the capacity and diversity of LMIC.

AI technologies should not encode biases to the disadvantage of identifiable groups, especially groups that are already marginalized. Bias is a threat to inclusiveness and equity, as it can result in a departure, often arbitrary, from equal treatment.

AI technologies should minimize inevitable disparities in power that arise between providers and patients, between policy-makers and people and between companies and governments that create and deploy AI technologies and those that use or rely on them. AI tools and systems should be monitored and evaluated to identify disproportionate effects on specific groups of people.

No technology, AI or otherwise, should sustain or worsen existing forms of bias and discrimination.

6. Promoting AI that is responsive and sustainable.

Responsiveness requires that designers, developers and users continuously, systematically and transparently assess AI applications during actual use. They should determine whether AI responds adequately and appropriately and according to communicated, legitimate expectations and requirements.

Responsiveness also requires that AI technologies be consistent with wider promotion of the sustainability of health systems, environments and workplaces. AI systems should be designed to minimize their environmental consequences and increase energy efficiency. That is, use of AI should be consistent with global efforts to reduce the impact of human beings on the Earth’s environment, ecosystems and climate.

Sustainability also requires governments and companies to address anticipated disruptions in the workplace, including training for health-care workers to adapt to the use of AI systems, and potential job losses due to use of automated systems.

A lightly edited Executive Summary from Ethics and Governance of Artificial Intelligence for Health by the World Health Organization.