• qa_deck_100#slide 1 · RAIDT core and mindmap orientation

RAIDT is a run-level evidence framework for responsible governance of generative AI systems used in organisational work. Across the papers, it is defined as a governance method rather than a single software product, and its practical outputs are a run-level evidence pack and a score profile for each material use. The framework makes governance readiness inspectable by scoring the five pillars (Responsibility, Auditability, Interpretability, Dependability, Traceability) and by treating recorded evidence, not narrative assurance, as the basis for review. In that sense, RAIDT operationalises the shift from high-level principles and model-level documentation towards reconstructable proof about what happened in context.

The mind map is therefore best organised with RAIDT as the root object and its core branches following the framework's own structure. A first branch covers Core and Scope: definition, purpose, contexts of use, and run as the unit of governance. Subsequent branches cover the required run-level evidence pack, the scoring logic and score profile, the anchors 1=missing / 3=partial / 5=audit-ready, boundary conditions and non-scope, and influence methods as governance interventions. A final branch captures worked vignettes and implementation pathways across healthcare, finance, public service, HR, education, and cybersecurity. This organisation mirrors the academic logic of RAIDT as an outcome taxonomy, an evidence protocol, and a scoring method, while keeping the whole framework conceptually prior to its internal parts.

Consider a public-service team using GenAI to interpret eligibility rules. In a RAIDT-centred mind map, the root node is not 'RAG' or 'auditability' but RAIDT itself. From that root, the analyst can move to Core and Scope to define the run, then to the run-level evidence pack to record the prompt, model version, retrieval snapshot, and checks, and then to the scoring branch to generate a score profile across the five pillars.

That structure matters operationally. If the rule text later changes or a citizen challenges the advice, reviewers do not begin with abstract policy claims; they begin with the recorded run and its evidence pack. The mind map therefore organises governance work around reconstructing and judging a concrete configured use, which is exactly what the papers identify as RAIDT's distinctive contribution.

• 00-RAIDT_Wording_v2
• 08-RAIDT_Foundations_M_V50
• 11-RAIDT_Academic_Logic_M_v11