S12.07 - Supervisor_reading_path

S12.07 ? Supervisor reading path

flowchart LR
    A[Fragmented reading of complex AI governance projects] --> B[RAIDT
Run-level evidence framework]
    A2[Component-first interpretation
policy first
example first] --> B
    B --> C[[Supervisor reading path
Centre-outward understanding]]
    C --> D[Run as unit of governance]
    D --> E[Evidence pack]
    E --> F[RAIDT score profile]
    C --> G[Reviewability and contestability]
    C --> H[Supervisory clarity and viva defence]
    C --> I[Policy and empirical work in correct sequence]
    J[Supervision meetings] --> C
    K[Workshop deck] --> C
    L[Paper planning] --> C
    M[Sector playbooks] --> C
    N[Governance dashboard] --> C
    F --> O[Governance readiness]
    E --> O

? Star S12 - Programme Architecture and Supervisory Navigation

Star context: This item defines the recommended centre-outward route for reading the RAIDT programme so that supervisors grasp the core framework first, then the run, the evidence model, the scoring logic, the interventions, the implementation pathways, and only then the wider empirical and policy extensions.

Academic picture

Definition / background

A supervisor reading path is the recommended order in which a supervisor, examiner, reviewer, or informed stakeholder should read the RAIDT project so that the conceptual centre is understood before the outer applications and extensions. In this item, the practical route is RAIDT -> Run -> Evidence pack -> Five pillars -> Interventions -> Implementation -> Policy -> Empirical programme.

Conceptually, this is an architectural device rather than a substantive governance artefact. It does not create evidence by itself, and it does not replace the framework. Instead, it protects the interpretation of the framework by ensuring that readers encounter the unit of governance, the evidence logic, and the scoring model in the right order. In complex doctoral work, this matters because a project can otherwise be mistaken for a set of papers, diagrams, domains, or tools rather than a coherent central proposition.

Within GenAI governance, the distinction matters because many governance discussions begin with abstract principles, isolated technical controls, or policy claims. RAIDT differs by treating the run as the unit of governance and by attaching evidence, scoring, and reviewability to that unit. The supervisor reading path belongs inside RAIDT because it helps readers see that the framework starts from run-level evidence and expands outward to interventions, implementations, and policy pathways.

The item therefore sits at the level of supervisory navigation and programme architecture. It links the core framework to the evidence pack, the RAIDT score profile, and the five pillars by showing how they should be understood as successive layers of one argument rather than as separate mini-projects.

Why this concept matters

Without a reading path, supervisors may enter the project through whichever component appears most visible at the time: a workshop deck, a policy chapter, an empirical example, or one pillar of the score profile. That creates a risk of category confusion. The framework can then appear fragmented, over-extended, or insufficiently bounded, even when the underlying logic is coherent.

The reading path solves that problem by making the sequencing explicit. It tells the reader that RAIDT is first a run-level evidence framework, then a model for organising evidence packs and score profiles, and only after that a basis for interventions, implementation choices, policy reasoning, and empirical validation. This prevents the outer rings from being mistaken for the centre.

For organisations using GenAI, the same logic matters operationally. If decision-makers begin with high-level principles or a generic policy checklist, they may never understand how evidence is actually generated and reviewed in practice. A centre-outward reading path helps move governance from slogans to inspectable, reviewable structure.

Key idea: The supervisor reading path matters because it preserves RAIDT's centre of gravity, ensuring that evidence, scoring, and governance claims are interpreted in the right conceptual order.

What this item enables

A disciplined centre-outward explanation of RAIDT for supervisors, examiners, and reviewers.
Clear separation between the core framework and its later extensions, examples, and outputs.
Faster supervisory orientation when time is limited and the whole project cannot be read at once.
Better coherence across papers, decks, viva answers, and policy-facing summaries.
Reduced risk that one component paper or implementation example is mistaken for the entire contribution.
A reusable route for teaching others how RAIDT moves from run-level evidence to governance readiness.

Practical example / likely audience question

Audience question

How can a supervisor understand RAIDT quickly without getting lost in the number of components, papers, diagrams, and application layers?

Answer

The concern behind the question is usually not lack of interest but lack of ordering. When a project contains a framework, a scoring model, intervention ideas, implementation pathways, policy implications, and empirical work, a supervisor may worry that the thesis has become too broad or that the parts do not belong together.

The direct answer is that RAIDT should be read from the centre outward. Start with RAIDT as the core run-level evidence framework. Then read the run as the unit of governance. After that, examine the evidence pack and the five-pillar score profile as the two principal outputs. Only once those are clear should the reader move to interventions, implementation pathways, policy implications, and the empirical programme.

A practical example is a supervision meeting in which the discussion begins with policy recommendations. That often leads to premature questions about national regulation, organisational mandates, or sector transfer. If instead the discussion begins with the run, the supervisor can first ask what is being governed, what evidence is collected, how the score profile is formed, and how reviewability is achieved. RAIDT handles this better than a generic AI governance approach because the reading path is anchored to a concrete unit of analysis and an evidence model, not merely to broad principles.

Practical example in RAIDT terms

Consider a university using a generative AI system to help academic staff draft student feedback. The specific run is one configured use of the system for a defined feedback task, at a particular time, by a named role, under institutional rules.

The run-level issue is that a supervisor or governance reviewer might jump straight to educational ethics or policy concerns without first understanding how RAIDT structures evidence. The reading path corrects that. It begins with RAIDT itself, then defines the run, then asks what evidence belongs in the evidence pack: task description, user role, prompt context, model configuration, human review expectations, logging records, quality checks, and escalation routes.

The affected pillars are Responsibility, because accountable use must be assigned; Auditability, because the run must be reconstructable; Interpretability, because reviewers need to understand what the system was doing; Dependability, because output quality and reliability matter; and Traceability, because the lineage of the run and its records must be visible. By using the reading path, a supervisor can see how the educational example is not an isolated case study but an application of the RAIDT governance logic, which improves governance readiness and presentation clarity.

Detailed link to RAIDT

Supervisor reading path links to RAIDT in four ways.

First, it protects RAIDT's core idea by ensuring that readers begin with the framework as a run-level evidence model rather than with a peripheral paper, domain example, or policy extension.

Second, it directs attention to the run as the unit of governance, which is essential because the rest of the project only makes sense once the governed unit is clear.

Third, it sequences the evidence pack and RAIDT score profile properly, helping readers understand how evidence is assembled before seeing how judgement and scoring are derived.

Fourth, it improves reviewability, contestability, audit readiness, and organisational learning because it teaches readers how to reconstruct the framework's reasoning in a stable order.

Supervisor reading path -> Run-level evidence -> Evidence pack -> RAIDT score profile -> Governance readiness

The item therefore acts as a bridge between programme architecture and governance logic. It does not add a new governance layer; it makes the existing governance architecture legible.

Link to the five RAIDT pillars

Responsibility

The reading path strengthens Responsibility by ensuring that governance questions begin with who is accountable for a run, what task is being undertaken, and what institutional purpose is being served.