Our AI Principles
Artificial intelligence (AI) can create positive, lasting change for our clients, communities, and businesses. At Catalyst we recognise the real-world impacts of AI and we are determined to build and use AI technology responsibly.
People first and human-centred
People are always in control of our AI systems, which we design to benefit people and minimise harm. For example, if an AI system’s decisions can have adverse impacts on people, there must be people who can review and override decisions.
We are committed to using AI to empower our colleagues and clients by enhancing their capabilities and supporting them to reach their goals.
Openness and transparency
We want to trust the technology we use and know what it’s doing. We do this through providing clear, auditable information proportionate to the risk of the system:
- outlining the likely impacts of risks
- clearly indicating when AI is in use
- describing in plain language what AI is doing
- indicating to people using a system enhanced by AI whether their data is used for AI training purposes.
Collaboration and partnership
We consult with clients, colleagues, and the wider community about the ethical use of AI tools. We invite people to be involved in the development of AI systems that use their data. If your project could benefit from AI, we discuss that thoroughly with you. We respect your knowledge, and we take your views seriously.
Respect and fairness
Any person or group that might be affected by the use of AI should be respected and given fair treatment. We use AI in a way that:
- supports self determination
- minimises discrimination or negative bias
- promotes inclusiveness.
Sovereignty and consent
We ensure people have control over their data, so our clients and AI system users:
- know where their data is stored and who has access to it
- are given an obvious, simple way to revoke access to their data.
Longevity and sustainability
The AI systems we work with provide lasting and positive benefits for our clients and staff, both now and in the future. A key priority for us is using environmentally sustainable methods to run AI systems. We use Catalyst Cloud, which has Toitū net carbonzero certification.
Reliability and security
We design AI systems with a focus on security and privacy for the systems’ entire life cycle. Ongoing monitoring ensures the systems continue to be reliable and dependable, and their outputs remain valid.
Open source
We apply open source principles and values in our AI work whenever it’s appropriate. For example, contributing code back to the community, prioritising the use of open source licences, and using open training data.
We partner with organisations that want to build and experiment with AI responsibly.
Whether that means developing local models that fit your specific needs, or hosting self-contained AI on NZ infrastructure so your data never leaves the country, we'll work alongside you to make it happen.