University of Pennsylvania Philadelphia, Pennsylvania
Abstract: Artificial Intelligence (AI) is quickly revolutionizing scientific research. While the tools that AI engenders purport to increase the rate of innovation and enable new predictive capabilities, they come with ethical challenges. This literature review examines discussions of AI ethics in genomics research over the past 25 years, mapping authors’ motivations for using AI methods and their discussion of potential ethical issues that come with it. AI has been deployed across a wide range of genomics research, such as interpreting genetic Variants of Unknown Significance (VUS), building drug-target prediction models for various cancer types, and developing AI-based multi- polygenic risk score (PRS) models that outperform non-AI single-PRS models. Proponents of incorporating these AI methods emphasize their ability to increase the speed and scalability of genetic techniques and usher in “a new era” of effective genomic medicine. Critics recognize the imperative to improve predictive genomics while cautioning against the overuse of AI in processes historically executed with human labor, noting the potential for algorithmic bias, deskilling, and a lack of transparency and explainability. Skeptics of the long term effects of AI cite the pressing need for ethical frameworks and regulation to guide its applications in genomics, forecasting risk to humans, other sentient beings, and the environment. By offering a proposed taxonomy of these challenges, this review provides a useful classification system for assessing the transformations and developments in approaches to AI ethics in genomics research.
Keywords: Artificial Intelligence, Genetics & Genomics, Ethical challenges in research innovation
Learning Objectives:
After participating in this conference, attendees should be able to:
At the end of the session, attendees will be able to assess arguments presented by proponents and critics of AI in genomics, weighing benefits of AI-driven innovation against ethical challenges.
At the end of the session, attendees will be able to recognize key ethical concerns related to AI in genomics research, such as algorithmic bias, deskilling, and lack of transparency.
At the end of the session, attendees will be able to evaluate ethical concerns about the implementation of AI in genomics research along temporal, disciplinary, and regulatory lines.
.png)