About the Project
Machine-human interaction is an essential component in technology law, including the recently adopted Artificial Intelligence Act (AI Act) in the European Union (EU). AI systems are required to be “sufficiently” and “appropriately” transparent to be overseen “effectively” by human beings.
The law is human-centric, as it is built on the premise that one or more individuals have legal obligations that benefit other humans. For example, doctors are legally obliged to provide appropriate science-based medical care, while patients have the right to receive such care.
The law will still be human-centric, irrespective of AI. Nevertheless, the prevalence of medical AI challenges conventional legal positions. The doctor is no longer only a doctor but also an operator of highly technical and complex automated systems. One doctor may potentially operate an AI system that affects an indefinite number of patients, which raises a series of interesting legal questions. Where should humans be placed in the lifecycle of an AI system, what should they be responsible for, and why? How far is the distance between the human patient and the human doctor? What types of medical AI can be offered without direct involvement from hospitals or local health services? May diagnosis and medical treatments be fully automated? If so, what does a fully automated treatment mean in legal terms?
Patient safety may be one justification for human involvement. An equally important justification may be patient autonomy, supported by a series of fundamental rights under the EU Charter of Fundamental Rights.
Explainable artificial intelligence (XAI) is an interdisciplinary research field, including computer science and social sciences, that research ways to make AI systems transparent and explainable to humans. This project is the legal component of a larger interdisciplinary project named RESPIRE. RESPIRE aims to create a global explainability framework for medical AI that includes computer science, medicine, ethics, and law perspectives. Hence, legal requirements will be connected to both the technical and social field of XAI.
Regulation of medical AI is not new. Medical devices that incorporate AI systems are approved under existing medical devices regulations. Furthermore, AI systems are regulated by data protection law, including the General Data Protection Regulation (GDPR). Combining old rules and the new Artificial Intelligence Act (AI) is highly complex. An essential part of the project is illuminating the connections between all the various regulations in a common explainability framework.
Goals
- Contribute the legal perspectives to a standard and interdisciplinary explainability framework.
- Connecting the bridge between legal and technical concepts, including transparency and explainability.
- Examining the nature, scope and justification for a legal right to human interaction with automated and autonomous technology, focusing on medical AI.
- Analysing limitations in regulating technology, including artificial intelligence.
Background
The project is based on a combination of doctrinal legal and critical legal analysis, including sociology and philosophy perspectives. The project relies on theories on the nature of law, the foundations of legal rights, analysis of legal positions, risk assessment and risk management theories, philosophy of technology and privacy, data protection and human autonomy.
Funding
The Norwegian Research Council funds the project as part of Respire - Responsible Explainable Machine Learning for Sleep-related Respiratory Disorders.
