Data Science Ethics

Current ethical concerns in data science related to the development and deployment of machine learning and artificial intelligence (AI) echo many concerns within the history of bioethics. We are at a critical juncture to learn from this history and avoid similar wrongs. Pressing concerns depend on philosophical analysis to determine how the ethical duties that we owe to others ought to guide the kinds of technology pursued. The potential benefit of a tool is another critical component of ethical analysis—and this is dependent on the tool’s explanatory value. Additionally, stakeholder perceptions are critical for determining an ethical path forward. I am in a unique position to contribute to the academic debates in this space given my expertise in bioethics, collaborations with data scientists, and ongoing participation in empirical research projects on applications of AI.

Project A – Bioethical Resources for Data Science

In our paper, “Ethics and Algorithms: Lessons from Public Health Ethics,” Suresh Venkatasubramanian, Kaitlin Pettit, and I argue that public health ethics offers an ideal format for guiding the development of ethics for data science because both areas require balancing between individual liberties against benefits for many, and a collective responsibility to protect the rights of others. Moreover, this framing highlights the importance of regulation as a manifestation of collective responsibility and the need for education highlighting the individual responsibilities of data scientists. Early versions of this argument have been presented at WCB 2020 and are currently being developed into a journal article (see Abstract 3A). This project emerged from a teaching collaboration with the computer science department, which produced the course Data Science Ethics, a collaborative pedagogical presentation, and I developed a pedagogical poster for the 2020 Teaching Hub at the annual Pacific Division Meeting of the American Philosophical Association.

Project B – Applications of AI and Machine Learning in Medical Decision Making

At present, I am involved in an empirical research project led by Jennifer Blumenthal-Barby and Kristin Kostick, in which we analyze stakeholders’ perceptions of the use of AI and machine learning to provide personalized risk predictions in LVAD decision making. The qualitative analysis sheds light on those elements that matter most to stakeholders including, patients, caregivers, and health care professionals. This project combines my interest in understanding, explanation, medical decision making, and applications of machine learning, and exemplifies my commitment to empirically informed and engaged philosophy.