S12.03 - Three-paper_arc

S12.03 ? Three-paper arc

flowchart LR
    A[Fragmented AI governance discussions] --> B[RAIDT - run-level evidence framework]
    A2[Single-paper overload] --> B
    B --> C[[Three-paper arc]]
    C --> C1[Paper 08 Foundations]
    C --> C2[Paper 09 Empirical Validation]
    C --> C3[Paper 10 Policy Pathways]
    C1 --> D[Run-level evidence]
    C2 --> E[Evidence pack and scoring credibility]
    C3 --> F[Policy alignment and interoperability]
    D --> G[Evidence pack]
    E --> H[RAIDT score profile]
    G --> I[Reviewer reconstruction]
    H --> J[Governance readiness]
    F --> J
    K[Healthcare and public services] --> C
    L[Finance and education] --> C
    M[Viva and supervisory explanation] --> C

? Star S12 - Programme Architecture and Supervisory Navigation

Star context: Helps supervisors understand how the RAIDT programme is organised across distinct but connected papers, so that the core framework, its validation, and its policy implications are not collapsed into one undifferentiated claim.

Academic picture

Definition / background

The three-paper arc is the programme-level research logic that structures RAIDT across three distinct but connected scholarly contributions. In this item, the arc consists of Paper 08 on foundations, Paper 09 on empirical validation, and Paper 10 on policy pathways. Together, these papers explain what RAIDT is, test whether its assessment logic can be used in a credible and repeatable way, and show how the framework can speak to wider governance and interoperability questions.

Conceptually, the item belongs to dissertation architecture rather than to a single RAIDT mechanism. Its role is to explain why the research is divided across papers with different functions instead of being compressed into one large text. In PhD terms, this matters because a framework paper must establish concepts and structure, a validation paper must examine whether the framework works in practice, and a policy paper must interpret what the findings mean for governance adoption, alignment, and institutional use. If these roles are blurred, the programme risks looking repetitive, methodologically unstable, or theoretically under-specified.

Within RAIDT, the three-paper arc matters because the framework is not only a conceptual model. RAIDT claims that the run is the unit of governance, that governance should be tied to evidence packs, and that performance and assurance should be represented through a five-pillar profile covering Responsibility, Auditability, Interpretability, Dependability, and Traceability. A three-paper structure provides room to define these ideas properly, evaluate them empirically, and then position them in the policy and organisational landscape.

This item differs from related concepts such as component papers or sector playbooks. Component papers may address specific elements within the broader programme, while sector playbooks adapt RAIDT to particular contexts. The three-paper arc, by contrast, explains the primary intellectual pathway of the core thesis contribution. It tells supervisors, examiners, and readers how the programme moves from theory to evidence to governance application without confusing implementation examples with the central academic claim.

Why this concept matters

The three-paper arc solves a common doctoral and governance problem: important ideas become unclear when conceptual design, empirical testing, and practical implications are mixed together without a visible structure. In AI governance research, this confusion often leads reviewers to ask whether the work is a framework, an evaluation study, a policy proposal, or a collection of examples. By making the arc explicit, RAIDT can answer that it is all of these in sequence, but not all at once in the same paper.

Without this item, there is a risk that the RAIDT programme appears over-claimed or under-disciplined. A reader may assume that the framework was never validated, that the validation lacks a conceptual basis, or that the policy discussion is detached from evidence. The arc prevents that fragmentation by showing that each paper has a distinct contribution and that each contribution supports the next.

For organisations using generative AI, this matters because governance maturity rarely emerges from principles alone. A framework must be defined, tested, and translated into pathways for adoption and oversight. The three-paper arc therefore supports the practical move from broad responsible-AI language to operational governance that can be reviewed, challenged, and improved.

Key idea: The three-paper arc matters because it turns RAIDT from a single thesis claim into a coherent programme in which conceptual foundations, empirical validation, and policy relevance are clearly separated and cumulatively connected.

What this item explains

How the RAIDT programme is organised across Papers 08, 09, and 10.
Why each paper has a different scholarly task and should not duplicate the others.
How conceptual foundations support empirical validation rather than compete with it.
How validation findings feed into policy pathways, interoperability, and governance adoption.
How supervisors and examiners can assess the coherence of the whole programme.
How the programme architecture protects the central RAIDT idea from being diluted by examples, tools, or sector-specific illustrations.

Practical example / likely audience question

Audience question

Why not put the full RAIDT framework, the validation study, and the policy argument into one major paper so that the contribution appears more complete?

Answer

The concern behind the question is understandable: reviewers often prefer a paper that appears self-contained, and doctoral programmes are sometimes pressured to present a single definitive output. However, combining framework construction, empirical validation, and policy interpretation in one paper usually weakens all three parts. The conceptual argument becomes rushed, the validation design becomes under-explained, and the policy discussion becomes speculative because it is not given enough analytical space.

In RAIDT, the direct answer is that each paper performs a different academic function. Paper 08 defines the run as the unit of governance, specifies the evidence pack, and establishes the five-pillar score profile. Paper 09 then asks whether that logic can be applied, assessed, and interpreted with sufficient consistency to support empirical claims. Paper 10 takes the framework and its validation outcomes into the governance environment, asking how RAIDT aligns with policy pathways, interoperability expectations, and adoption contexts.

A practical example is a reviewer asking whether Traceability scores in RAIDT have any real evidential value. If that question appears in the foundations paper, the answer is conceptual: the paper explains why traceability matters and how it fits the framework. If it appears in the validation paper, the answer is empirical: the paper shows whether assessors can use the scoring logic consistently. If it appears in the policy paper, the answer is institutional: the paper shows how such traceability could support audit readiness or governance alignment. RAIDT handles this better than generic AI governance approaches because it does not stop at principles. It separates design, validation, and governance translation in a way that makes each claim reviewable.

Practical example in RAIDT terms

Consider a public-service use case in which a generative AI system helps staff draft responses to citizen complaints. The organisation wants assurance that the system is being used responsibly, that outputs can be reviewed after the fact, and that decisions influenced by the system remain contestable.

The run-level issue is that each drafting event is context-specific: the prompt, policy guidance, source material, user role, timing, and resulting output may differ from one case to another. RAIDT therefore requires run-level evidence rather than general claims about the tool. The evidence needed includes the run configuration, prompt and instruction context, user role, artefact history, model choice, output record, review notes, and pillar-level scoring justification.

The three-paper arc improves governance readiness because it shows how this use case would be treated across the programme. Paper 08 explains why a complaint-response event should be governed at run level and how the evidence pack and five pillars structure the assessment. Paper 09 validates whether reviewers can use the scoring and evidence logic reliably in practice. Paper 10 then asks how such run-level evidence could fit organisational policy, accountability processes, and broader public-sector governance expectations.

The most affected RAIDT pillars in this example are Responsibility, Auditability, and Traceability, though Interpretability and Dependability also matter. The arc matters because it shows that the framework is not merely proposed, but examined and then translated into governance pathways that a public organisation could plausibly adopt.

Detailed link to RAIDT

Three-paper arc links to RAIDT in four ways.

First, it protects the RAIDT core idea by making clear that the project is centred on run-level evidence for governing generative AI in organisational work, not on a loose bundle of separate themes.
Second, it shows how the run becomes academically legible across the programme: defined in foundations, examined in validation, and interpreted in governance terms in policy pathways.
Third, it connects the evidence pack and score profile to a cumulative research design, so these outputs are not introduced as isolated artefacts but as objects that are conceptualised, tested, and positioned for use.
Fourth, it strengthens reviewability, contestability, audit readiness, and organisational learning because each stage of the programme can be inspected for a distinct purpose and then related back to the overall governance claim.

Three-paper arc ? Run-level evidence ? Evidence pack ? RAIDT score profile ? Governance readiness

The chain matters because RAIDT is strongest when its architecture is visible: the programme explains the logic of the run, turns that logic into assessable evidence, and then shows how that evidence supports practical governance.

Link to the five RAIDT pillars

Responsibility

The three-paper arc supports Responsibility by clarifying where normative claims are made, where they are tested, and where they are translated into governance expectations. This avoids vague responsibility language that is never operationalised.