Responsibility Infrastructure.
The architecture for responsibility that must remain intelligible, provable and usable across organisational and system boundaries.
Definition
Responsibility Infrastructure provides the architecture through which responsibility can be represented, evidenced, verified, recognised, resolved and relied upon across independent organisations and systems.
Its purpose is to preserve responsibility as it moves across boundaries, so that an independent party can determine what responsibility existed, who or what held it, under what authority, what occurred, what evidence supports the record and what standing applies at the point of reliance.
Responsibility Infrastructure Standard
The Responsibility Infrastructure Standard is a title used for the existing RI-originated technical requirements. Each publication has its own stated scope and version; no externally approved standard or mandatory hierarchy is implied.
Responsibility Infrastructure Standard overview →Working proposals for information exchange
Published working material examines the responsibility information and exchange behaviour required for independent systems to preserve meaning across boundaries.
Where expressly released, the public package may include schemas, reference code, validators, fixtures and conformance tests necessary for independent interoperability. Internal implementation details that are not part of the applicable public specification remain outside interoperability conformance.
Exchange working proposal →Normative scope
Existing RI documents describe distinct functions within the wider architecture. They do not add undisclosed obligations to a separately agreed interoperability specification. Any RI publication claim must identify its version, role and scope.
Functions
Responsibility records preserve what happened. Verification determines what the evidence supports. Recognition determines effect within the applicable recognition regime. Registry preserves and publishes standing. Resolution determines which standing applies to a particular request. Reliance occurs when an external party or system uses that standing within its own decision context.
Implementation neutrality
The public interoperability layer does not prescribe a vendor, universal software architecture, database, user interface or deployment model. Independent implementations may use different internal systems provided the applicable interoperability requirements are satisfied and responsibility meaning survives the boundary.
Application
AI assurance is one application of Responsibility Infrastructure; Responsibility Infrastructure is not limited to AI. It can apply wherever responsibility crosses organisational, institutional or system boundaries across people, organisations and machines.
It can be applied wherever responsibility crosses organisational, institutional or system boundaries, including human-to-human, human-to-machine, machine-to-human and machine-to-machine contexts.