Stated Position History in AI-Assisted Decisions: What It Preserves, What It Does Not Establish, and Evidence of Reuse Across Two Domains

Records of human involvement in AI-assisted decisions are increasingly read as evidence that a person examined the matter and checked the result. This paper separates three things that such a reading merges: the fact of involvement, the participant's stated position, and understanding. The subject of this paper is the stated position history: a retained sequence of records of what a participant stated about a consequential decision — what was considered, what remained uncertain, and what position was taken — together with the provenance of each recorded element. Not legal advice. Not a compliance certification. This paper presents methodological proposals, not established legal doctrine.