2,000+ interaction cycles.
A six-month private autoethnographic archive of prompts, responses, manuscript revisions, and reflective notes.
Public research release · Complete paper
Reflexive
Authorship
Generative AI as Algorithmic Mirror and the Reconfiguration of Creative Agency.
The model proposes. The author governs.
OPEN ACCESSComplete revised edition · 15-page PDF
The complete revised paper is public. Read it online, download the PDF, and cite the edition with attribution.
The governing proposition
Participation may be distributed.
Accountability remains concentrated.
01 · The conceptual turn
Not a tool.
Not a second mind.
Reflexive Authorship treats generative AI as an apparatus of return: language is transformed, returned, judged, and allowed to reshape the next human intention.
intention → return →
reorientation
- Cognitive loop exposure
- Repetition externalizes the author’s patterns, making signature and mannerism newly distinguishable.
- Reflexive co-agency
- The model changes the cost and visibility of possible formulations; the human retains purpose and responsibility.
- Algorithmic intra-action
- Some concepts emerge across a sequence of prompt, variation, rejection, revision, and accountable closure.
The framework does not attribute consciousness, intention, moral status, or legal authorship to a model.
02 · Evidence discipline
The polish
cannot testify.
The revised edition separates what the corpus can illuminate from what it cannot prove.
Concept-generating evidence.
The corpus supports a situated theoretical account. It does not establish population-level psychological effects or causal universality.
Prose cannot testify for evidence.
Every claim must stand on its sources, limits, and capacity to be checked—not on the polish of its language.
Governed execution.
Future empirical work requires timestamped sequences, model records, a disclosed codebook, comparative cases, and independent review.
The abstract, publicly rendered
The author
after return.
Generative AI is commonly described either as a tool that extends human intention or as a collaborator that contributes to creative production. Both descriptions miss a consequential feature of sustained interaction: model output returns to the user as an externalized and probabilistic transformation of linguistic habits, conceptual preferences, and unresolved assumptions. Reflexive Authorship names the recursive process through which that return alters the next prompt, the next judgment, and eventually the author’s account of creative agency.
03 · Four refusals
What this edition
does not claim.
That generative AI possesses consciousness, intention, or moral agency.
That a model glitch reveals subconscious truth without human interpretation and documented context.
That one author’s longitudinal experience automatically generalizes across users, languages, or systems.
That a fluent manuscript earns epistemic authority merely by sounding complete.
04 · Public release
The position is public.
So is the paper.
The original 53-page working bundle has been rebuilt as a clean 15-page research edition. The complete paper—including the named three-part framework, corpus analysis, evidence protocol, repaired references, objections, and authorship standard—is now publicly available.
- Author
- Tony T. Nguyen | Nguyễn Từ Vĩnh Trường
- Publisher
- Highland Reflexive Press
- Record
- HRP–RN–008
- Form
- Public research release
- Length
- 15-page typeset PDF
- Status
- Independent · not peer reviewed
- Access
- Open PDF
- Citation
- Attribution required
This public release may be read, downloaded, and cited with attribution. Redistribution as a substitute edition, republication, model training, adaptation, institutional course distribution, and commercial use require prior written permission. No copyright, publication, or commercial right is transferred.