• integrated_82#Q4.14

The ageing-society calibration adds three sector moderators to RAIDT: vulnerability, inclusion, and ability to challenge decisions. These do not replace the core RAIDT structure; rather, they calibrate what acceptable evidence and acceptable performance look like when services affect older adults. Vulnerability lowers tolerated error and raises expectations for uncertainty signalling, safeguarding, and human handover. Inclusion turns Interpretability into a socio-technical requirement, so explanations must be usable by older adults, carers, and frontline staff rather than merely technically plausible. Ability to challenge decisions shifts attention from generic transparency to contestability, meaning that reasons, policy bases, and appeal routes must be visible in the record.

The calibration also adds operational artefacts. The paper translates the moderators into ageing-service requirements such as accessible explanations, uncertainty and escalation, ability to challenge decisions, safeguarding awareness, audit-ready records, privacy and dignity, change control, and interoperability. In implementation, these requirements are assessed through the run-level evidence pack and a sector-calibrated score profile rather than through policy claims alone. The healthcare playbook makes this operational with anchors 1=missing / 3=partial / 5=audit-ready and by specifying the evidence fields another reviewer needs to reconstruct the run. In short, the ageing-society calibration adds boundary conditions, evidence duties, and review criteria that make RAIDT workable for high-consequence services involving older adults.

A local authority uses a GenAI assistant to explain whether an older adult qualifies for home-support transport. Under the ageing-society calibration, the output cannot simply state a result in confident prose. The system must provide plain-language reasons, the policy basis used, and a clear appeal route. If RAG is used, the run-level evidence pack stores the retrieved rule text, its identifier, and the retrieval date so the advice can be reproduced later.

This addition changes governance in a practical way. A reviewer can inspect the score profile and see whether Interpretability was achieved for the citizen, whether Auditability and Traceability support challenge, and whether Responsibility required escalation to a human caseworker. If the evidence only partially supports reconstruction or the appeal steps are absent, the run should sit at anchors 1=missing / 3=partial / 5=audit-ready rather than be treated as deployment-ready.

• 20-RAIDT_AgeingSoc_M_V50
• 21-RAIDT_Sector_Playbook_Healthcare_V2