S12.02 - Core_governance_artefacts

S12.02 ? Core governance artefacts

flowchart LR
    A[Fragmented AI governance
inconsistent terminology
ad hoc review] --> B[RAIDT
run-level evidence framework]
    B --> C[[Core governance artefacts]]
    C --> C1[Canonical wording]
    C --> C2[Scoring rubric]
    C --> C3[Evidence pack structure]
    C --> C4[Prompt and influence blueprints]
    C --> C5[Run-level templates]
    C --> D[Evidence pack]
    C --> E[RAIDT score profile]
    D --> F[Reviewer reconstruction]
    D --> G[Traceable governance record]
    E --> H[Five-pillar judgement]
    H --> I[Governance readiness]
    F --> I
    G --> I
    J[Healthcare]
    K[Finance]
    L[Public services]
    M[Education]
    N[Enterprise productivity]
    J --> C
    K --> C
    L --> C
    M --> C
    N --> C

? Star S12 - Programme Architecture and Supervisory Navigation

Star context: Helps supervisors understand the whole RAIDT programme without confusing components, papers, sector examples, or implementation details with the small set of governance artefacts that keep the framework conceptually stable and operationally reusable.

Academic picture

Definition / background

Core governance artefacts are the reusable formal objects that hold RAIDT together as a coherent framework rather than a loose collection of ideas. In this item, the term refers to the canonical wording used to define RAIDT consistently, the scoring rubric used to interpret run-level evidence, the evidence pack structure used to organise what is collected, the prompt and influence governance blueprints used to frame control over model behaviour, and the run-level templates used to standardise documentation and review.

Conceptually, these artefacts matter because governance frameworks often fail at the point where abstract principles must be translated into repeatable practice. Many organisations can state that AI should be responsible, auditable, interpretable, dependable, and traceable, but they lack durable artefacts that tell reviewers what to look for, how to compare runs, and how to record findings in a form that supports scrutiny. RAIDT addresses this translation problem by making the artefacts themselves part of the governance design.

This item differs from a generic discussion of documentation or policy. A policy statement expresses intent. A governance artefact structures action, evidence, and review. Within RAIDT, the artefacts are not peripheral administrative materials; they are the mechanisms through which run-level evidence becomes legible, comparable, and scorable. They are therefore directly linked to both practical outputs of the framework: the run-level evidence pack and the five-pillar score profile.

Because RAIDT treats the run as the unit of governance, the artefacts must operate at that same level. They help define what counts as relevant evidence for a single run, how that evidence should be interpreted, and how it should feed into a Responsibility, Auditability, Interpretability, Dependability, and Traceability assessment. In that sense, core governance artefacts are part of RAIDT's architecture of reviewability.

Why this concept matters

Core governance artefacts solve a coordination problem inside AI governance. Without them, different reviewers may describe the same run differently, apply inconsistent standards, or produce evidence that cannot be meaningfully compared over time. This weakens audit readiness, frustrates supervisory understanding, and makes it difficult to defend the programme in a viva, paper, or organisational review.

The concept also prevents a frequent confusion in doctoral and practice settings: the belief that the RAIDT programme is held together mainly by its papers, examples, or sectors. Those outputs matter, but they are not the core binding mechanism. What actually holds the programme together is the shared governance machinery that keeps each run interpretable through the same logic of evidence, scoring, and review.

If these artefacts are missing, organisations may still perform governance activities, but those activities remain ad hoc. The result is governance by assertion rather than governance by structured evidence. RAIDT is designed precisely to move beyond that condition by embedding artefacts that make scrutiny operational.

Key idea: Core governance artefacts matter because they convert RAIDT from a set of principles into a repeatable run-level governance system.

What this item enables

It enables consistent wording of RAIDT concepts across papers, supervision discussions, and implementation materials.
It enables standardised collection and arrangement of run-level evidence in a reusable evidence pack.
It enables scoring consistency by giving reviewers a common rubric for interpreting evidence across the five pillars.
It enables prompt and influence governance to be examined through structured blueprints rather than informal judgement.
It enables comparison across runs, sectors, and case studies without collapsing important contextual differences.
It enables supervisor and examiner navigation by showing that the programme has stable governance components beneath its various outputs.
It enables organisational learning, because evidence from one run can inform improvement in later runs through shared templates and review structures.

Practical example / likely audience question

Audience question

What holds the programme together?

Answer

The concern behind this question is usually that RAIDT may appear to be an umbrella label covering several papers, examples, and governance ideas without a sufficiently stable centre. The direct answer is that the programme is held together by reusable governance artefacts: the canonical wording, scoring rubric, evidence pack structure, prompt and influence governance blueprints, and run-level templates.

A practical way to see this is to imagine two different RAIDT applications: one in healthcare and one in enterprise productivity. The domains differ, the tasks differ, and the risks differ. Yet both can still be governed through the same underlying artefacts. Each run can be documented through a common template, assessed through the same five-pillar rubric, and organised into an evidence pack that a reviewer can inspect. That continuity is what keeps RAIDT coherent.

A generic AI governance approach might answer this question by pointing to principles, policies, or committee structures. RAIDT answers it more concretely. It identifies the specific artefacts that make evidence collection, scoring, review, and comparison possible at run level. That is why the framework remains stable even when sectors, tasks, or implementation contexts vary.

Practical example in RAIDT terms

Consider a healthcare organisation using a generative AI assistant to draft discharge-summary text for clinicians. The use case is operationally valuable, but the run-level issue is that a specific output may have been shaped by particular prompt instructions, model settings, contextual data, and reviewer interventions at a specific moment. If these details are not captured consistently, the organisation cannot later explain why a summary was produced, whether it met quality expectations, or how governance controls were applied.

Core governance artefacts improve this situation by supplying a standard run template, a defined evidence-pack structure, and a scoring rubric for assessing the run. The evidence needed would include the configured task definition, prompt form, model version, user role, contextual constraints, reviewer checks, output record, and rationale for any overrides or escalation. The most affected RAIDT pillars would be Responsibility, Auditability, Dependability, and Traceability, with Interpretability also relevant where explanation of output logic is needed.

Because the artefacts specify how evidence is framed and reviewed, governance readiness improves. A supervisor, auditor, or clinical governance lead can reconstruct the run more confidently, compare it with other runs, and identify where improvements to process or controls are required. The artefacts therefore make the use case governable in practice rather than merely approved in principle.

Detailed link to RAIDT

Core governance artefacts link to RAIDT in four ways.

First, they operationalise RAIDT's core idea that governance should be organised around evidence from a specific run rather than around abstract claims about a system in general.
Second, they shape how the run is documented, interpreted, and reviewed at the level of actual use.
Third, they provide the structure through which evidence is assembled into the evidence pack and translated into a RAIDT score profile.
Fourth, they support reviewability, contestability, audit readiness, and organisational learning by making governance judgements reconstructable and comparable.

Core governance artefacts ? Run-level templates and review rules ? Evidence pack ? RAIDT score profile ? Governance readiness

This chain matters because RAIDT is strongest when the movement from run to evidence to score to governance action is explicit. Core governance artefacts are what keep that movement stable.

Link to the five RAIDT pillars

Responsibility

Core governance artefacts clarify who is expected to define, review, approve, and act at each point in the governance process. They reduce ambiguity about roles and obligations.