AUGMANITAI Core Vocabulary, version 1.0: Bilingual Terms for Human Oversight and for Situations between People, AI Systems and Software Agents

Abstract

Release 1.3 of the AUGMANITAI core vocabulary: 121 bilingual (English/German) terms. Release 1.3 adds the first terms of two MANITAI dimensions: BRAMANITAI, experience at the brain–computer interface (17 terms based on an analysis of published first-person accounts, for example setup weight, hold weariness, unmeant output, handed start, effort fade, day-to-day feel of control), and ENTMANITAI, artificial entities as counterparts (19 terms, for example parallel counterpart, entity boundary blur, self-description gap, unverified mandate, persona residue, successor attribution, fluid entity, intent anchor, pull to the mean). Terms of earlier releases are unchanged; their frozen version 1 stays as published. New in this release: each new term links to its MANITAI dimension (dct:subject, https://w3id.org/augmanitai/manitai//; register: https://doi.org/10.5281/zenodo.23411689), and a first set of SKOS mappings to Wikidata in SSSOM, Turtle and JSON-LD connects terms to established concepts (closeMatch, broadMatch, relatedMatch; no exactMatch). Every term has a persistent identifier under https://w3id.org/augmanitai/lexicon/ and a frozen version identifier ( /v1). Each entry gives a definition, a delimitation from neighbouring terms, a plain-language reading, a reading for language models and software agents (rules, record fields and a testable condition), notes for standards bodies and organisations, and a citation. Formats: JSON-LD (schema.org DefinedTermSet and SKOS), Turtle (SKOS), JSON Lines (one term per line). A SHA-256 manifest lists every file. Licence: CC BY 4.0. Attribution: Andreas Ehstand, ORCID 0009-0006-3773-7796, https://w3id.org/augmanitai