A New Frontier of Ethics Oversight: Preclinical Data-Generation at ASBH Conference
October 15 @ 11:45 am – 12:45 pm
Hastings Center for Bioethics Associate Scholar Julia Kolak will present a paper at the 2026 annual ASBH Conference.
Abstract: Bioethics scholarship on functional genomics and artificial intelligence (AI) has largely concentrated on downstream issues arising in clinical translation, model deployment, and patient-facing applications. By contrast, comparatively little attention has been paid to ethical, legal, and social issues (ELSIs) embedded upstream in the design of AI-ready biomedical datasets. Drawing on insights from an embedded ethics initiative within a major AI-oriented genomics data project, this presentation examines whether prevailing ELSI frameworks and research oversight structures are adequately positioned to address the normative implications of preclinical research at the emerging frontier of AI-ready biomedical datasets.
Three recurring patterns illustrate central challenges of integrating ethics into early-stage scientific workflows.
First, responsibility for anticipating the downstream implications of data generation choices is often diffuse across the research ecosystem, leaving unclear who bears accountability for design optimization and the values those choices may incorporate or overlook. Second, investigators frequently struggle to map familiar ethical categories onto highly technical decisions in experimental design and data curation, and are often uncertain how to prioritize the value-laden trade-offs embedded in everyday workflows. Third, researchers lack shared frameworks for ethical reflection, while ethics teams may lack the domain-specific technical insight needed to guide these decisions.
Together, these findings highlight critical limits in existing research oversight structures. Because many ethically consequential decisions shaping AI-ready datasets occur outside domains typically governed by human subjects review, addressing this gap may require updated research ethics frameworks and closer integration of domain-specific expertise to evaluate the ethical stakes of upstream data practices.
