Preserving Claim Revisions in Records of AI-Assisted Philosophical Inquiry

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

Philosophical claims can change through dialogue, source encounters, objections, and editorial work. A final article presents the current argument, while leaving much of that development outside its scope. This paper develops a proposed documentary architecture for preserving consequential claim revisions in AI-assisted philosophical inquiry. It distinguishes inquiry events, available traces, chronological records, analytical commentary, and stabilized articles. The architecture retains three functions: preserving the documented course of inquiry, interpreting its development, and presenting the currently defended argument. Selected antecedents in Plato's dialogue framing, scholarly text encoding, provenance representation, and oral-history ethics clarify the proposal's scope. A minimal typed graph separates temporal order, documentary derivation, and argumentative relations. A finite construction shows how selective release can remove an indirect documentary connection even when its endpoints remain public; a separate counterexample distinguishes recorded sequence from intellectual causation. The analysis derives proportionate responsibilities for fidelity, disclosure, and correction from explicit normative premises, with privacy and legitimate limits on preservation included. Hashes, stable identifiers, and complete exports contribute through specific functions while leaving event truth, conceptual priority, and philosophical validity in need of separate warrants. The proposed architecture supports future scrutiny and continuation under declared limits. Its practical usability, costs, and institutional implementation remain open to investigation.