Transparency is the degree to which an AI system’s purpose, data use, operation, limitations, and outputs can be understood or examined by relevant stakeholders. It supports accountability, informed use, auditing, and governance, but does not always require full disclosure of every internal parameter.
‘property of a system that appropriate information about the system is made available to relevant stakeholders’ ISO/IEC 22989.
Making the operation and decision-making processes of AI systems clear and understandable to users and stakeholders. Key components of transparency are: Openness: Clearly communicating the purpose and capabilities of an AI system. This includes explaining what the system is designed to do and any limitations it may have. Explainability: Providing understandable explanations of how the AI system reaches it decisions. Accountability: Ensuring that there's a mechanism for tracking and verifying decisions made by the AI. This can include maintaining logs, version control and audit trails. Data Transparency: Disclosing what data is used to train and operate the AI system, including its sources and how its processed.