When applied in the health sector, AI-based applications raise not only ethical but legal and safety concerns, where algorithms trained on data from majority populations can generate less accurate or reliable results for minorities and other disadvantaged groups.

In October 2021, the Allen Institute for Artificial Intelligence publicly released Delphi, an artificial intelligence system trained to make general moral decisions (Allen Institute for Artifi...

Artificial intelligence surveillance can be used to diagnose individual cases, track the spread of Covid‐19, and help provide care. The use of AI for surveillance purposes (such as detecting new Covid‐19 cases and gathering data from healthy and ill individuals) in a pandemic raises multiple concerns ranging from privacy to discrimination to access to care. Luckily, there exist several frameworks that can help guide stakeholders, especially physicians but also AI developers and public health officials, as they navigate these treacherous shoals. (...) While these frameworks were not explicitly designed for AI surveillance during a pandemic, they can be adapted to help address concerns regarding privacy, human rights, and due process and equality. In a time where the rapid implementation of all tools available is critical to ending a pandemic, physicians, public health officials, and technology companies should understand the criteria for the ethical implementation of AI surveillance. (shrink)

Employers and governments are interested in the use of serological testing to allow people to return to work before there is a vaccine for SARS-CoV-2. We articulate the preconditions needed for the implementation of antibody testing, including the role of the U.S. Food & Drug Administration.

Artificial Intelligence (AI), especially of the machine learning (ML) variety, is used by health care organizations to assist with a number of tasks, including diagnosing patients and optimizing operational workflows. AI products already proliferate the health care market, with usage increasing as the technology matures. Although AI may potentially revolutionize health care, the use of AI in health settings also leads to risks ranging from violating patient privacy to implementing a biased algorithm. This chapter begins with a broad overview of (...) health care AI and how it is currently used. We then adopt a “lifecycle” approach to discussing issues with health care AI. We start by discussing the legal and ethical issues pertaining to how data to build AI are gathered in health care settings, focusing on privacy. Next, we turn to issues in algorithm development, especially algorithmic bias. We then discuss AI deployment to treat patients, focusing on informed consent. Finally, we will discuss existing oversight mechanisms for health AI in the United States: liability and regulation. (shrink)