The Journal of Prompt-Engineered (Moral) Philosophy Or: Why AI-Assisted Ethics Research Requires Process Transparency

Michele Loi

arxiv: 2511.08639 · v3 · pith:Y44REIFOnew · submitted 2025-11-10 · 💻 cs.CY · cs.AI· cs.DL

The Journal of Prompt-Engineered (Moral) Philosophy Or: Why AI-Assisted Ethics Research Requires Process Transparency

Michele Loi This is my paper

Pith reviewed 2026-05-21 19:49 UTC · model grok-4.3

classification 💻 cs.CY cs.AIcs.DL

keywords AI disclosureethics researchprocess transparencyagent integrityphilosophical methodologymeaningful human controldocumentation framework

0 comments

The pith

Transparency in AI-assisted ethics research serves agent integrity by making the author's philosophical commitments legible to their community.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper contends that existing rules for disclosing AI use in scholarship fall short because they do not explain the purpose of transparency. Ethical inquiry is contested both over its proper subject matter and over what it requires of the person doing the inquiring, which rules out judging such work solely by its outputs or by economic criteria borrowed from empirical fields. Instead, transparency is required to preserve agent integrity: the ability of a community to see the identity-shaping commitments expressed in how the author philosophizes. The achievable aim is therefore not consensus evaluation but tracking, so that each tradition can later assess the work on its own terms. To make this concrete, the paper introduces and demonstrates a documentation-adequacy framework built around five transparency elements.

Core claim

The transparency duty is grounded in agent-integrity: the legibility, before a community of inquiry, of the identity-constituting commitments that the author's mode of philosophising expresses. Because standards for evaluating philosophical work are not communally settled, the goal of transparency is tracking rather than evaluation against agreed criteria, thereby accumulating an evidentiary record that lets future normative judgments remain possible.

What carries the argument

A documentation-adequacy framework with five elements—declaration, navigation, documentation account, process documentation, and development records—that operationalizes Meaningful Human Control.

If this is right

Reproducibility standards imported from empirical sciences do not apply to ethics research because the relevant criteria are not settled in advance.
The goal of transparency shifts from enabling external judgment against fixed rules to enabling each philosophical tradition to track and evaluate the work on its own later terms.
Meaningful Human Control over AI-assisted inquiry is achieved through the five-element documentation framework rather than through output inspection alone.
Future normative assessments of the work become possible only if an archival record of the process is preserved and accessible.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The same tracking approach could be tested in other contested domains such as political theory or applied ethics where community standards also remain open.
Journals might need to require persistent process archives as a condition of submission rather than simple disclosure statements.
The framework could be extended by asking authors to record how specific AI prompts interacted with their own prior commitments.

Load-bearing premise

Ethical inquiry is essentially contested at two independent levels about what it is and about what it demands of the inquirer.

What would settle it

A case in which output-only review or welfare-economic criteria alone suffice to assess the legitimacy of AI-assisted philosophical claims without any record of the author's process or commitments.

read the original abstract

Existing AI disclosure mandates in scholarship require that AI assistance be reported but leave transparency philosophically unspecified: they fix the duty without explaining what the duty serves. We argue that ethical inquiry is essentially contested at two independent levels -- about what it is, and about what it demands of the inquirer -- defeating output-only evaluation and welfare-economic dismissal of the transparency question, and, by extension, reproducibility framings imported from the empirical sciences. The transparency duty is grounded instead in agent-integrity: the legibility, before a community of inquiry, of the identity-constituting commitments that the author's mode of philosophising expresses. Because the standards for evaluating such work are not communally settled, the achievable goal for transparency is not evaluation against agreed criteria but tracking -- accumulating the evidentiary record that lets each tradition assess the work on its own terms and makes future normative judgments possible. We develop a documentation-adequacy framework that operationalises Meaningful Human Control through five transparency elements -- declaration, navigation, documentation account, process documentation, and development records -- demonstrated by the paper itself, whose full documentation record is archived at a persistent identifier. The framework is a first iteration subject to revision, not a settled standard.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

Loi grounds AI disclosure in ethics research in agent-integrity rather than reproducibility and supplies a five-element documentation framework as a concrete first pass.

read the letter

The main point is that this paper treats transparency for AI use in philosophical ethics as a matter of keeping the author's own commitments legible to the community, not as a reproducibility or output-evaluation exercise. It starts from the claim that ethical inquiry is contested both about what it is and about what it requires of the inquirer, which blocks welfare-economic dismissals and imported empirical standards. From there it develops a documentation-adequacy framework with five elements—declaration, navigation, documentation account, process documentation, and development records—to operationalize meaningful human control and enable tracking across traditions. The paper applies the framework to its own process and points to an archived record, which gives the proposal some practical grounding that most disclosure mandates lack. That combination of the agent-integrity rationale and the specific five-element list is the clearest new piece relative to earlier AI disclosure work. The soft spot is the missing step that would show why these particular elements are necessary or sufficient for rendering identity-constituting commitments legible. The paper asserts the operationalization but does not derive the list from the contestation premise or explain how each element avoids collapsing back into a general reproducibility concern. A tighter linkage would make the framework less detachable from the central argument. This is for philosophers and AI-ethics researchers who care about process norms in conceptual scholarship. A reader looking for a starting proposal on how to document AI assistance without pretending the standards are settled would get value from it. I would send it to peer review so the framework and its grounding can be tested against alternatives.

Referee Report

1 major / 1 minor

Summary. The manuscript argues that existing AI disclosure mandates in scholarship are philosophically underspecified because they fix a transparency duty without explaining its purpose. It claims that ethical inquiry is essentially contested at two independent levels—about what ethical inquiry is and what it demands of the inquirer—which defeats output-only evaluation, welfare-economic dismissals, and reproducibility framings imported from the empirical sciences. Transparency is instead grounded in agent-integrity: the legibility, before a community of inquiry, of the identity-constituting commitments expressed by the author's mode of philosophising. The paper develops a documentation-adequacy framework that operationalises Meaningful Human Control through five transparency elements (declaration, navigation, documentation account, process documentation, and development records), demonstrated by the paper itself via an archived full documentation record at a persistent identifier. The framework is presented as a first iteration open to revision.

Significance. If the central argument holds, the work supplies a philosophically grounded alternative to empirical reproducibility standards for AI-assisted ethics research, emphasising evidentiary tracking that accommodates unsettled communal standards rather than evaluation against agreed criteria. The self-demonstration via an archived persistent-identifier record is a concrete strength that models the proposed process transparency and could influence disclosure practices in philosophy and adjacent fields.

major comments (1)

[Section developing the documentation-adequacy framework] The grounding of the transparency duty in agent-integrity (legibility of identity-constituting commitments before a community) is not explicitly derived from the essential-contestation premise. The manuscript asserts that the five elements operationalise Meaningful Human Control to achieve this legibility, but supplies no argument showing why declaration, navigation, documentation account, process documentation, and development records are necessary or sufficient for rendering commitments legible, nor how they follow from contestation rather than from a separate reproducibility concern. This risks rendering the framework detachable from the integrity rationale.

minor comments (1)

[Abstract] The abstract could more explicitly flag that the framework is offered as an initial, revisable proposal rather than a settled standard, to manage reader expectations about its scope.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their constructive engagement and recommendation for major revision. The comment identifies a valuable opportunity to strengthen the explicit linkage between the essential contestation premise and the documentation-adequacy framework. We address this below and will revise accordingly.

read point-by-point responses

Referee: [Section developing the documentation-adequacy framework] The grounding of the transparency duty in agent-integrity (legibility of identity-constituting commitments before a community) is not explicitly derived from the essential-contestation premise. The manuscript asserts that the five elements operationalise Meaningful Human Control to achieve this legibility, but supplies no argument showing why declaration, navigation, documentation account, process documentation, and development records are necessary or sufficient for rendering commitments legible, nor how they follow from contestation rather than from a separate reproducibility concern. This risks rendering the framework detachable from the integrity rationale.

Authors: We agree that the derivation could be rendered more explicit to foreclose any impression of detachment. The manuscript grounds transparency in agent-integrity precisely because two-level essential contestation (about the nature of ethical inquiry and what it demands of the inquirer) renders communal evaluative standards unsettled; this defeats output-only assessment and reproducibility framings that presuppose fixed criteria. The five elements are therefore presented as the minimal operationalisation of Meaningful Human Control that enables evidentiary tracking rather than evaluation: declaration makes AI involvement legible so that authorship commitments can be assessed; navigation orients readers within the author's reasoning path; the documentation account articulates the rationale for methodological choices; process documentation records iterative decisions; and development records preserve the full history. These follow from contestation because they accumulate the record that permits each tradition to judge on its own terms without requiring shared standards. We will add a dedicated subsection immediately preceding the framework presentation that derives each element step-by-step from the contestation argument, demonstrating necessity for legibility of identity-constituting commitments and sufficiency for tracking without invoking empirical reproducibility. This will tighten the integration and address the concern directly. revision: yes

Circularity Check

0 steps flagged

No significant circularity: philosophical premises remain independent of the proposed framework.

full rationale

The paper derives the transparency duty from premises about essential contestation in ethical inquiry and grounds it in the independent concept of agent-integrity (legibility of identity-constituting commitments). It then introduces the five-element documentation framework as an operationalization of Meaningful Human Control, demonstrated by the paper itself, without any self-definitional reduction, fitted parameters renamed as predictions, or load-bearing self-citations that collapse the argument back into its inputs. The central claim does not equate the conclusion to a quantity or definition constructed from the framework elements; the derivation is self-contained against external philosophical benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The paper draws on domain assumptions about the contested character of ethical inquiry and the limits of output-only assessment; no free parameters or invented entities are introduced.

axioms (1)

domain assumption Ethical inquiry is essentially contested at two independent levels -- about what it is, and about what it demands of the inquirer.
Invoked to defeat output-only evaluation and welfare-economic framings of transparency.

pith-pipeline@v0.9.0 · 5746 in / 1170 out tokens · 60653 ms · 2026-05-21T19:49:17.782956+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Foundation/RealityFromDistinction.lean reality_from_one_distinction unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

The transparency duty is grounded instead in agent-integrity: the legibility, before a community of inquiry, of the identity-constituting commitments that the author's mode of philosophising expresses.
IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

five transparency elements — declaration, navigation, documentation account, process documentation, and development records

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

Epistemic Constitutionalism Or: how to avoid coherence bias
cs.AI 2026-01 conditional novelty 7.0

AI systems require an explicit epistemic constitution with procedural norms to avoid coherence biases such as source attribution sensitivity.

Reference graph

Works this paper leans on

4 extracted references · 4 canonical work pages · cited by 1 Pith paper

[1]

Begin with SP-3 (Reproduction Guide) — human-directed instructions explaining the overall workflow, comparison criteria (trajectory matching), pass threshold, and the old/new mapping (see A.4)

work page
[2]

Provide the LLM with three inputs: o SP-1 (Complete Prompt) o SP-2.1 (Reproduction Package) — processed extracts with Architectural Overview, Guidance & Refinement Patterns, Source Integration, Development Flow, Key Insights o SP-2.2 (Reproduction Procedure) — LLM-directed instructions for how to use SP-1 + SP-2.1 together to generate the reproduction

work page
[3]

undocumented work)

Compare generated work to submitted paper following criteria in SP-3: intellectual architecture match, presence of key insights, reproducibility of major moves, and gap analysis (expected editorial refinement vs. undocumented work)

work page
[4]

Use the mapping table (A.4) to translate old Roman labels to new Arabic section numbers

If needed, consult SP-4 for deeper investigation. Use the mapping table (A.4) to translate old Roman labels to new Arabic section numbers. For Editorial Assessment • SP-1 shows disclosed inputs (foundational instructions). • SP-2 provides a processed overview of scope and depth. • SP-3 offers a clear verification procedure. • SP-4 demonstrates maximal tra...

work page

[1] [1]

Begin with SP-3 (Reproduction Guide) — human-directed instructions explaining the overall workflow, comparison criteria (trajectory matching), pass threshold, and the old/new mapping (see A.4)

work page

[2] [2]

Provide the LLM with three inputs: o SP-1 (Complete Prompt) o SP-2.1 (Reproduction Package) — processed extracts with Architectural Overview, Guidance & Refinement Patterns, Source Integration, Development Flow, Key Insights o SP-2.2 (Reproduction Procedure) — LLM-directed instructions for how to use SP-1 + SP-2.1 together to generate the reproduction

work page

[3] [3]

undocumented work)

Compare generated work to submitted paper following criteria in SP-3: intellectual architecture match, presence of key insights, reproducibility of major moves, and gap analysis (expected editorial refinement vs. undocumented work)

work page

[4] [4]

Use the mapping table (A.4) to translate old Roman labels to new Arabic section numbers

If needed, consult SP-4 for deeper investigation. Use the mapping table (A.4) to translate old Roman labels to new Arabic section numbers. For Editorial Assessment • SP-1 shows disclosed inputs (foundational instructions). • SP-2 provides a processed overview of scope and depth. • SP-3 offers a clear verification procedure. • SP-4 demonstrates maximal tra...

work page