John Fox has spent most of his career working on the theoretical foundations for and practical delivery of AI systems in healthcare (Safe and Sound: Artificial Intelligence in Hazardous Applications, AAAI and MIT Press, 2000). This experience has yielded a general approach to the design of "intelligent software agents” which are capable of using human-like knowledge models in automated reasoning, decision-making, planning and management of patient care. The CREDO design framework has proved to be effective in a wide range of clinical applications, with 20 successful trials to date, most of which have been reported in peer reviewed medical and technical journals. Such evidence suggests that well established AI technologies already have the ability to help healthcare professionals improve the quality and safety of patient care and reduce costs in many areas of medical practice. Furthermore AI technologies are continuing to evolve (e.g. machine learning, knowledge representation) and many new capabilities will be demonstrated in the coming years, including autonomous systems which support many elements of routine patient care. This of course raises a vast range of non-technical issues in which the skills of social scientists and legal specialists will be needed.
Prof Fox will raise a number of questions that have already been encountered in medical AI for discussion with the members of the Oxford Law and AI group, including:
· What is the duty of care on those who design and deploy AI services, particularly where patient safety is a concern?
· How transparent and accountable must AI systems be? (the famous “black box” problem)
· How can we apportion professional responsibility and legal liability where quality of service is in dispute?
· We live in an accelerating “knowledge crisis”: how do we maintain AI services as professional views of best practice evolve?
· What is the legal framework within which autonomous medical agents should be introduced and regulated? (e.g. “smart contracts”)
· Can different legal and ethical systems be aligned across jurisdictions and cultures?
· How can engineers and lawyers collaborate effectively to address these questions?