AI Life-cycle Accountability
Roles across the AI ecosystem
Overview
The first step toward better governance and goal attainment is to establish accountability.
The Structured Accountability Workbench treats clear accountability as the foundation of effective AI governance. It examines the roles and responsibilities in development and deployment that influence decision-making and whether an AI system achieves its intended purpose.
Effective performance, responsible stewardship and ethical behaviour depend on precise roles. The workbench presents a layered view of the AI ecosystem, showing the level and scope of accountability for the organisation's role - provider, deployer or operator - and the corresponding duties of its governing body.
Governance of decision-making is part of overall organisational governance. Authority is delegated so that work can be shared, but AI decision-making policy must keep humans clearly accountable for the authority they have delegated.
The governing body monitors the decisions and outputs of automated systems and directs management to keep those systems within acceptable bounds. It also seeks assurance that oversight is assigned to staff who are resourced and authorised to act when issues arise.
Accountability is shown through accurate, timely reporting on performance and stewardship of resources. The governing body retains and distributes value transparently and reports on processes, decisions, results and organisational impact over time.
Accountability for the governance of AI remains with the governing body, regardless of any delegation.
