AI in Medicine Between Technology, Ethics and Medical Responsibility

Why the German Medical Association’s Publication “Medical Practice with Artificial Intelligence” Is Worth Reading

Author: Dr Eveline Prochaska, Dresden University of Technology, 30 January 2026

Artificial Intelligence (AI) has long become part of medical practice – often operating invisibly in the background, sometimes the subject of controversy. The question is no longer whether AI should be used in medicine, but rather how responsibly, how transparently and within what limits.

With its publication Medical Practice with Artificial Intelligence (May 2025), the German Medical Association (Bundesärztekammer, BÄK) presents, for the first time, a comprehensive position paper from a medical perspective that systematically brings together technology, clinical practice, ethics and regulation [1]. For anyone working with AI in medicine – whether in research, clinical practice, teaching or industry – this publication offers several particularly relevant perspectives.

A central theme of the publication is a clear distinction:

AI should support medical practice, not replace it. Diagnosis, treatment decisions and therapy remain explicitly the responsibility of healthcare professionals, even as AI systems become increasingly capable [1].

This position is particularly relevant in the context of modern Clinical Decision Support Systems (CDSS) and Large Language Models (LLMs), which are capable of aggregating, structuring and generating suggestions from complex clinical information. The German Medical Association emphasises that these systems must remain support tools, including the requirement that physicians assess the plausibility of their outputs [1].

A second focus lies on the role of medical data. The publication makes it clear that AI in medicine is not possible without standardised, interoperable and high-quality data [1].

The following are explicitly highlighted:

• Clinical routine data from hospitals
• Registry data
• Claims and reimbursement data
• Data from the electronic patient record (ePA)

The German Medical Association also refers to existing national infrastructures such as the Medical Informatics Initiative (MII), the Network University Medicine (NUM), the National Research Data Infrastructure for Personal Health Data (NFDI4Health) and federated computing approaches, all of which make AI applications possible in the first place [1].

The publication becomes particularly tangible when describing specific fields of application, including:

• Medical imaging (radiology, pathology and dermatology)
• Intensive care and emergency medicine (monitoring and prognostic models)
• Cross-sector healthcare
• Outpatient care and telemedicine
• Documentation and clinical workflows
• Medical research (e.g. protein structure prediction)

It becomes clear that many AI applications are already in use today, often without being recognised as such [1].

A distinguishing feature of the publication is its in-depth ethical and legal perspective. The Central Ethics Committee of the German Medical Association highlights, among other things:

• Responsibility remains with humans
• Trust in systems is a challenge at both the meso and macro levels
• AI must not replace medical empathy and the physician–patient relationship
• Transparency, validation and quality assurance are essential requirements [1]

This perspective complements international discussions on Trustworthy AI and regulatory requirements such as the European Union’s AI Act [2].

Publication Download / Link

Anyone seriously engaged with AI in medicine – whether from a strategic, scientific, or practical perspective – will find:

• Clear positions
• Realistic application scenarios
• Well-founded ethical guidance
• A bridge between medicine, computer science, and regulation