# Expert System **Domain:** Artificial Intelligence / Knowledge Representation **Doc Type:** Canonical Concept Node **Maturity:** Foundational ## Definition An **expert system** is a program that applies explicitly represented domain knowledge and inference procedures to problems ordinarily requiring specialized human judgment. ## Historical Context [[wiki/DENDRAL|DENDRAL]] is widely treated as the first expert system for scientific hypothesis formation; [[wiki/MYCIN|MYCIN]] applied rule-based consultation to antimicrobial therapy. Both were rooted in biological or medical domains at Stanford. [[articles/The Evolutionary Roots of Silicon Valley|The Evolutionary Roots of Silicon Valley]] uses this history to reverse the claim that AI merely borrowed metaphors from biology. Biologists and clinicians helped build foundational AI systems because their fields already confronted inference under incomplete evidence. ## Architecture and Limits Expert systems separate a reasoning engine from encoded domain knowledge, but the separation is never complete. Representation choices determine what the engine can notice. Systems can perform strongly within a bounded domain while lacking common sense, temporal flexibility or institutional authority. ## Governance Context Capability does not resolve custody. Clinical use requires responsibility for errors, update procedures, explanation, workflow integration and a human authority able to override recommendations. ## Sources / Provenance - Stanford Biomedical Data Science history: https://dbds.stanford.edu/biomedical-data-science-graduate-program-overview/ - Feigenbaum and Buchanan, DENDRAL historical papers: https://cs.stanford.edu/people/eaf/wordpress/cv/ ## See Also [[wiki/DENDRAL|DENDRAL]], [[wiki/MYCIN|MYCIN]], [[wiki/Reasoning Engine|Reasoning Engine]], [[wiki/Encoded Domain Knowledge|Encoded Domain Knowledge]], [[wiki/Moral Accountability Layer|Moral Accountability Layer]]