Safeguard human rights from automated Artificial Intelligence decision-making

What the Topic Is

This petition asks Parliament to stop using Artificial Intelligence (AI) or automated systems to make decisions about welfare benefits. "Automated decision-making" means a computer program calculates whether someone gets money, housing, or support, without a human checking the result. The petition specifically targets decisions that suspend, reduce, or end benefits. It cites "Robodebt," a controversial Australian system that incorrectly accused people of owing money, as a warning of what can go wrong.

Current Situation in New Zealand

New Zealand’s welfare system, managed primarily by Work and Income (part of the Ministry of Business, Innovation and Employment, or MBIE), uses digital tools to process claims. While the government uses data matching to detect errors or fraud, it currently relies on human caseworkers to make final decisions on benefit entitlements. There is no specific legislation in NZ that explicitly bans AI from making these final calls, but current practice involves human oversight. The petition exists because some citizens fear that as technology advances, the line between "assisting" a worker and "deciding" for them may blur, removing human accountability.

Arguments in Favour

• Human Dignity: Supporters argue that vulnerable people deserve empathy and individual consideration, which algorithms cannot provide.
• Transparency: It is often difficult to understand how an AI reaches a conclusion ("black box" problem), making it hard for people to appeal decisions.
• Bias and Error: Automated systems can inherit biases from their training data or make calculation errors that disproportionately harm low-income individuals.

Arguments Against

• Efficiency and Speed: Opponents argue that automation helps process large volumes of applications quickly, reducing wait times for those who genuinely need help.
• Consistency: Algorithms apply the same rules to everyone, potentially reducing human bias or inconsistent treatment by different caseworkers.
• Resource Constraints: The government may argue that fully human review is too costly and slow for the current scale of the welfare system.

Relevant Context

The primary legal framework for welfare in NZ is the Social Security Act 2018. This law outlines how benefits are administered but does not explicitly address the use of AI in decision-making. The "Robodebt" scandal in Australia (2016–2020) is a key reference point; an independent review found the system was unfair and caused significant distress. In NZ, the Office of the Privacy Commissioner and the Human Rights Commission have raised concerns about data privacy and algorithmic fairness in public services. The people most affected are those receiving benefits such as the Jobseeker Allowance, Supported Living Payment, or Family Tax Credit.

⚡ Cached from a previous visit