Help & support
Artificial Intelligence (AI) is developing rapidly. Appropriate use of AI will help us to achieve our purpose of building a brighter future for all. We recognise that a failure to govern and manage AI systems effectively may lead to undesirable consequences such as incorrect decision-making, compliance failures and poor customer outcomes. We aim to address these risks through our governance and compliance program.
We have a principles based approach to the policies governing the design, development, deployment and use of AI systems by the Bank.
The six AI principles are:
Our AI policy may intersect with the Bank’s other policies, as relevant. For example, there may be specific Bank policies where the AI system is also a model or where the AI system is sourced from a supplier.
We view AI as a machine based system that independently learns from data, designed to operate with varying levels of autonomy and that can, for explicit or implicit objectives, generate outputs such as content, predictions, recommendations, or decisions influencing the environments they interact with. AI includes technologies such as machine learning, Generative AI (identifies patterns and relationships in data, including supervised, unsupervised and reinforcement learning), dynamic or adaptive models, speech recognition, natural language processing and computer image recognition.
The following six AI Principles are embedded within the processes and governance which are applicable to design, development, deployment and use of AI systems by the Bank:
AI systems should advance human, social and environmental wellbeing as well as facilitate respect for human rights, diversity, and the autonomy of individuals. To justify the balance of potential harms and benefits that an AI system delivers means that other solutions, including not deploying any system, have been considered and ruled-out because they do not realise the same overall benefits delivered by the AI system.
All people should be treated fairly and must not be unfairly discriminated against. Where AI systems contribute to decisions and outcomes, those decisions and outcomes are judged against the same standards that would apply to decisions and outcomes made entirely by humans.
AI systems should be transparent so that their manner of operating and outputs can be readily understood, reproducible and, where appropriate, contested. Reference resources must be written in plain language and at an appropriate level of detail.
AI systems must ensure security of data and comply with privacy and data protection laws, including in accordance with the Bank’s privacy and security policies.
AI systems should operate reliably, perform consistently and in accordance with their intended purpose. AI systems should not pose unreasonable safety risks, and should adopt safety measures that are proportionate to the potential risks.
Human oversight of AI systems is necessary. There must be sufficient oversight by individuals with relevant expertise in the technology, the intended use, benefits and risks relevant to the AI system.
