Precedent-Based Professional Role Ethics for AI Decision Analysis

Large language models (LLMs) are increasingly being used in professional fields such as healthcare, law, and engineering. In these domains, errors can lead to serious consequences. Many current AI ethics approaches do not reflect the structured codes and precedent-informed reasoning that guide professional conduct. This work introduces ProEthica, a system under development that combines LLMs with role-based ontologies to support structured ethical reasoning in professional settings. The system draws on principles from professional role ethics and incorporates ethical guidelines, practice standards, and prior case decisions. As a demonstration case, it applies the National Society of Professional Engineers (NSPE) Code of Ethics and Board of Ethical Review precedents. ProEthica includes an ontology based on engineering ethics, a precedent retrieval method using both vector similarity and ontological mappings, a framework for guiding and checking LLM outputs with structured constraints, and a validation process modeled on FIRAC (Facts, Issues, Rules, Analysis, Conclusion) reasoning. The system is intended to help professionals make ethical decisions that are consistent with established standards, not to replace human judgment. Preliminary evaluations using NSPE cases indicate that it can retrieve relevant precedents and produce structured analyses that align with engineering ethics.