Watson, autonomy and value flexibility: revisiting the debate

Area of Science:

• Bioethics
• Medical Informatics
• Artificial Intelligence in Healthcare

Background:

• Ethical concerns surrounding Clinical Decision Support Systems (CDSSs) are significant.
• Previous discussions have focused on the impact of IBM's Watson for Oncology (Watson) on patient autonomy.
• Autonomy, responsibility, fairness, and transparency are key ethical considerations for CDSSs.

Purpose of the Study:

• To elaborate on the impact of Watson on patient autonomy.
• To propose conceptual, empirical, and technical methods for evaluating and incorporating autonomy in CDSS design.
• To clarify the relationship between value-flexible design and value-sensitive design.

Main Methods:

• Utilizing Jonathan Pugh's account of rational autonomy to analyze decisional and practical autonomy.
• Drawing an analogy with patient decision aids to develop an empirical measure of autonomy (value-congruence).
• Situating technical options for integrating autonomy into Watson's design within the framework of value-sensitive design.

Main Results:

• Watson's presentation of results may affect decisional autonomy, while its knowledge production methods may impact practical autonomy.
• Value-congruence, analogous to patient decision aids, offers an empirical method to estimate Watson's impact on autonomy.
• Value-flexible design can be clarified and situated within value-sensitive design, offering technical solutions for incorporating patient autonomy.

Conclusions:

• The study provides a tripartite framework (conceptual, empirical, technical) for understanding and enhancing patient autonomy in the context of AI-driven CDSSs like Watson.
• It offers practical tools and design considerations for developing more ethically aligned and patient-centered AI in healthcare.
• Addressing autonomy is crucial for the responsible development and deployment of advanced medical technologies.