S1.01 - Original_PhD_route

S1.01 ? Original PhD route

flowchart LR
    A[Responsible AI
managerial uncertainty
information disorder] --> B[RAIDT
run-level evidence framework]
    H[Practical settings
healthcare, finance, public services
prompts, sources, review notes] --> C[[Original PhD route
from broad governance problem
to operational RAIDT design]]
    I[IS governance
audit and accountability lineage
GenAI operational pressure] --> B
    B --> C
    C --> D[Run-level evidence pack]
    C --> E[RAIDT score profile]
    D --> F[Reviewer reconstruction
contestability]
    E --> G[Governance readiness
organisational learning]

? Star S1 - Origins, Background and History

Star context: Explains the intellectual route by which RAIDT emerged from Responsible AI, managerial uncertainty, information disorder, IS governance, audit traditions, and GenAI operational pressure, rather than from a single narrow concern such as auditability alone.

Academic picture

Definition / background

The original PhD route refers to the initial conceptual path from which RAIDT emerged. The project did not begin as a narrowly technical study of audit logs, nor as a purely abstract discussion of AI ethics. It began with a broader Responsible AI concern: how management and organisational actors can use AI-supported decision processes under conditions of uncertainty, incomplete information, misinformation, dynamic change, and accountability pressure. From the start, the problem was practical as well as conceptual. Organisations needed ways to justify decisions and review AI-assisted work when the informational environment was unstable and when outputs could not be treated as fully deterministic or self-explanatory.

This route matters because it explains why RAIDT eventually took its present form. Responsible AI supplied the normative concern. Managerial uncertainty supplied the decision context. Information disorder highlighted the difficulty of relying on contested or incomplete inputs. IS governance contributed the organisational and control perspective. Audit and accountability traditions contributed the expectation that actions should later be reconstructable and reviewable. GenAI operational pressure then made the issue urgent, because organisations were already using systems whose outputs were probabilistic, configurable, and often woven into everyday work.

Within RAIDT, the original PhD route therefore functions as a design rationale. It explains why the framework is centred on run-level evidence rather than broad policy claims alone, why the evidence pack is necessary, and why the five-pillar score profile combines Responsibility, Auditability, Interpretability, Dependability, and Traceability. These are not arbitrary elements added after the fact. They are the operational expression of the original research problem.

The concept also differs from a generic project history. In RAIDT, the original route is analytically important because it clarifies the framework's scope. RAIDT is about responsible governance of real GenAI use in organisational work. It belongs neither to pure model evaluation nor to generic ethics rhetoric alone. Its intellectual route explains why the framework sits between governance theory, operational assurance, and evidence-based review of concrete runs.

Why this concept matters

This concept matters because it prevents RAIDT from being misunderstood as a framework that emerged only from auditability or only from compliance thinking. If the original route is forgotten, the framework can appear narrower than it really is. In reality, RAIDT addresses a more difficult problem: how to govern situated uses of GenAI when people must act under uncertainty and still remain answerable for outcomes.

The concept also avoids confusion between a research origin story and a governance logic. The origin route is not merely autobiographical background. It explains why RAIDT moves from principles to operational evidence, why it focuses on the run, and why the framework treats governance as something that must work in practice rather than only in policy documents. Without this item, audiences may miss the intellectual coherence linking Responsible AI, organisational governance, and run-level review.

For organisations using GenAI, this matters because governance frameworks are often challenged on two fronts at once: they are criticised for being too abstract to operationalise and too technical to explain to decision-makers. The original PhD route shows why RAIDT is neither. It begins from an organisational problem of accountable decision support and arrives at a practical governance method grounded in evidence packs and score profiles.

Key idea: The original PhD route matters because it explains why RAIDT treats GenAI governance as accountable decision support under uncertainty, operationalised through run-level evidence rather than left at the level of principle alone.

What this item explains

The intellectual and practical origins of RAIDT as a response to Responsible AI concerns in organisational decision-making.
Why managerial uncertainty and information disorder were central to the project before the framework was formalised.
How IS governance and audit traditions helped shift the project from broad ethical concern to structured organisational accountability.
Why GenAI operational pressure made the move from conceptual governance to run-level evidence urgent.
Why RAIDT uses the run as the unit of governance rather than relying only on model-level or policy-level assurance.
How the evidence pack and five-pillar score profile emerged from the original governance problem rather than from an arbitrary design preference.
Why RAIDT should be understood as an operational governance framework for organisational work, not only as an audit device or an ethics vocabulary.

Practical example / likely audience question

Audience question

Did RAIDT start as auditability only?

Answer

No. Auditability became one important pillar, but it was not the sole or original framing of the PhD route. The earlier concern was broader: how to support responsible organisational decisions when AI is used under uncertainty, when information may be incomplete or misleading, and when actors remain accountable for the consequences of using those outputs.

The misconception behind the question is that RAIDT began as a technical assurance mechanism and then expanded outward. The actual route was almost the reverse. The project began with a broad governance and decision-support problem, then narrowed toward the practical question of what kind of evidence would allow a specific use of GenAI to be reviewed, contested, and justified. Auditability was therefore a consequence of the original route, not its only source.

A practical example is a manager using GenAI to help draft a strategic risk briefing from mixed-quality internal and external information. The problem is not merely whether the interaction was logged. The deeper issue is whether the organisation can show what information was used, what the tool produced, how uncertainty was handled, who reviewed the result, and whether the outcome was governance-ready. RAIDT handles this better than a generic AI governance approach because it ties that practical problem to the run, to evidence capture, and to a structured scoring profile rather than stopping at high-level policy statements.

Practical example in RAIDT terms

Consider a public-service team using GenAI to draft an internal briefing on housing-support demand for the next quarter. The use case seems straightforward, but the underlying information is uncertain: some data are current, some are delayed, and some narrative inputs come from informal frontline reports that may be incomplete or inconsistent.

The run-level issue is not only whether the GenAI system produced fluent text. The issue is whether the briefing can be trusted as a governance input. Evidence is needed on the task definition, the prompt or template used, the source datasets and notes supplied, the model configuration, the generated draft, human edits, reviewer comments, and the final version used in decision-making. Responsibility is affected because roles and sign-off must be clear. Auditability is affected because the run must be reconstructable. Interpretability is affected because reviewers need to understand how the output related to the inputs and instructions. Dependability is affected because unstable inputs may produce misleading summaries. Traceability is affected because the briefing must be linked to time, source material, and downstream use.

This item improves governance readiness because it explains why such evidence is needed in the first place. The original PhD route frames the run not as an isolated technical event but as a situated organisational decision-support episode under uncertainty. That framing is what makes RAIDT especially useful for operational governance.

Detailed link to RAIDT

Original PhD route links to RAIDT in four ways.

First, it explains the RAIDT core idea that GenAI governance should begin from real organisational use under uncertainty, not only from abstract principle statements or model-level claims.

Second, it leads directly to the run as the unit of governance, because the original research problem was how to examine one concrete episode of AI-supported work in context.

Third, it explains why RAIDT needs both a run-level evidence pack and a five-pillar score profile: evidence is needed for reconstruction, while scoring is needed for structured judgement across key governance dimensions.

Fourth, it connects RAIDT to reviewability, contestability, audit readiness, and organisational learning by showing that each of these depends on understanding how a particular run was shaped, used, and assessed.

Original PhD route ? Run-level evidence ? Evidence pack ? RAIDT score profile ? Governance readiness

The item therefore acts as a bridge between the conceptual origins of the project and the practical mechanics of the framework. It shows that RAIDT's architecture is a response to a governance problem that began before the framework had a name.

Link to the five RAIDT pillars

Responsibility

The original route strongly influences Responsibility because the project began from the question of how organisational actors remain answerable when GenAI contributes to decisions or work products under uncertainty.