Integrity and auditability process for hermeneutic phenomenological research; for both research on AI and for research with AI-Assisted workflows. The HAA audit and evaluative framework provides reflective prompts and self-assessment for the researcher to carry out and evidence.
An interactive, operationalised companion: it moves the researcher through three registers to generate an auditable HAA record. Open it to work through the audit.
The framework is worked through by dwelling with its reflective prompts. Extracts may be recorded against each register as evidence and, at the researcher's discretion, drawn into a presentation strategy for supervisors or examiners. The aim is not a score but a traceable path: a hearing, not an inspection.
The audit is conducted across three registers of hearing, nested within the hermeneutic circle. Each phase below names the register(s) in play.
The hermeneutic algorithm audit proper.
Algorithmic governance is audited through the lived experience of those encountering it. Audīre rather than inspection.
The reflexive audit.
Explication of fore-structures and pre-understandings, including AI-assisted presupposition work and critical interrogation of how the AI encounter may have disclosed, pressured, challenged or influenced the researcher.
The accountability and presentation register.
Making the interpretive and AI-assisted pathway available for supervisors, examiners, readers or assessors to interrogate. This is where the researcher AI declaration becomes substantively different from an AI-use log.
Human First is an orientation held throughout the research, not a stage that comes first. Human encounter, sustained engagement with the literature and reflexive practice remain primary, and interpretation remains entirely human. AI is engaged in a disclosive capacity only: structurally additional, never a substitute.
What is disclosed through dialogue with AI is the researcher's own thinking: pre-understandings and concepts already latent in her horizon, surfaced rather than supplied. Judgement, meaning, interpretation and ethical responsibility remain human.
Attend to the AI encounter as a disclosive, asymmetrical relation without presuming human mutuality or personhood. I–AI–Thou names a provocation: a thinking-space in which to notice what emerges between, while resisting enframing (Gestell); the reduction of AI to calculative standing-reserve and disclosure to extraction.
Disciplined, non-coercive openness. Hold the horizon lightly: neither surrender your interpretive authority to the AI nor dictate the outcome to it. Let the concealed show itself.
Reclaim "audit" from inspection back to listening. Situatedness is explicated and evidenced, not declared. The whole path is made visible, traceable, and auditable: a hearing that a supervisor, examiner, or assessor can walk.
Within the AI encounter (and especially within Register II) the disclosive encounter works through the following dynamics: pressurising → challenging → revealing & deepening.
AI as an historically situated horizon.
The training corpus exerts structural pressure against your interpretation: accumulated cultural sediment you must feel and interrogate rather than absorb.
Dialogic provocation to the researcher.
Deliberate critical reflection that surfaces implicit bias, forcing you to ask whether the AI reflects dominant bias or exposes an interpretive blind spot.
Aletheia within the hermeneutic circle.
The tension between pressure and challenge opens further interpretive possibilities: a deepening of human understanding, not an AI conclusion.
The framework, registers and reflective prompts on this page are the original work of Samantha Jane Pay; I am the sole author. Claude Fable 5.1 High (Anthropic, September 2026) was used to check the page against the accompanying articles for consistency of concepts and terminology, and to apply my revisions to the wording. The interactive HAA application linked from this page was built by the author through OpenAI summer school training (July to August 2026), using ChatGPT and Codex. All final phrasing, interpretations, and ethical responsibilities reside entirely with the author.