The Green Party has pledged to prevent artificial intelligence and automated decision-making systems from making decisions that cancel, suspend or reduce welfare payments if it enters government.
The proposal would require consequential decisions about financial support to remain subject to human involvement. It adds another significant AI governance issue to New Zealand’s election debate, following Labour’s recently announced wider AI Action Plan.
Key facts
- The proposal covers decisions to cancel, suspend or reduce welfare payments.
- It would also cover automated decisions imposing benefit sanctions.
- The Greens say people affected by welfare decisions should be able to deal directly with a person.
- The policy responds in part to concerns about automated decision-making at the Ministry of Social Development.
- The announcement follows growing international scrutiny of automated government decision systems.
Our take
There is an important distinction between using AI to help public servants make decisions and allowing automated systems to make consequential decisions themselves.
For New Zealand organisations, the broader lesson applies well beyond government. Where AI influences employment, lending, insurance, healthcare or other decisions that materially affect people, organisations should define where human judgement remains mandatory and who is accountable for the final decision.
Sources
KiwiGen.AI 