Before AI Learns Traditional Chinese Medicine: Cosmological Grounding and Negotiated Implementation
This preprint examines a fundamental question in the application of artificial intelligence (AI) to traditional Chinese medicine (TCM): when AI systems reproduce diagnostic features, classifications, or reasoning patterns, which conception of TCM are they actually representing, and whose interpretive assumptions guide their outputs? Drawing on historical scholarship, recent developments in TCM-related AI, and previous conceptual work on question framing and the semantic meaning layer–cosmological meaning layer (SML–CML) framework, this paper proposes cosmological grounding as an approach to making interpretive and implementation decisions inspectable, contestable, and revisable. The central proposal consists of six interconnected disclosure domains: Conceptual provenance Tradition selection and interpretive priority Question framing and cosmological commitments Contemporary constraints and conflicts Negotiated implementation Accountability and revisability These domains are accompanied by illustrative documentation requirements and two hypothetical applications: an AI-assisted pattern differentiation system incorporating different TCM traditions, and a commercial yangsheng (life cultivation) application. The framework does not privilege a single interpretation of Chinese medicine. It can also accommodate AI systems developed around particular medical schools, lineages, or individual practitioners, provided that the sources and grounds for their claims of fidelity remain open to examination. The paper distinguishes computational performance, theoretical coherence, and clinical validity. It does not claim to validate traditional diagnostic theories or demonstrate clinical effectiveness. Rather, it argues that responsible AI implementation requires making visible what knowledge and values were selected, what was modified or excluded, why those decisions were made, who is accountable, and how they may be challenged or revised. This is a conceptual preprint and has not undergone peer review.
Authors
- Kimio Shiraishi (ORCID: https://orcid.org/0009-0003-2550-7385)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-21
- DOI
- https://doi.org/10.5281/zenodo.24414027
- Primary Topic
- Artificial Intelligence in Healthcare and Education
- Type
- preprint