Politicians Bet on AI to Solve Australia's Housing Crisis — at Risk of Another Robodebt Tragedy

A New Era in Housing with AI

Paul Scully, the New South Wales Minister for Planning and Public Spaces, has called the state government's initiative to launch a tender for an artificial intelligence (AI) solution to the housing crisis "a game changer." This move is aimed at reducing bureaucratic hurdles and accelerating the construction of more homes. The system is expected to be fully operational by the end of 2025.

Scully emphasized that this approach would allow construction projects to begin and new homeowners to move into their properties faster. The federal treasurer, Jim Chalmers, also supported the initiative, calling it a model for other states and territories to follow in unlocking more housing and boosting economic productivity.

Speeding up building approvals is a central focus of the so-called abundance agenda, which aims to stimulate economic growth. Other regions in Australia are already taking steps in this direction. Tasmania is developing an AI policy, while South Australia is trialing a small-scale pilot for specific dwelling applications, allowing users to submit digital architectural drawings for automatic assessment against set criteria.

Will AI Be the Quick Fix?

Despite the enthusiasm around AI, questions remain about its effectiveness in addressing Australia’s housing crisis. Cutting red tape is a key theme in discussions about improving productivity. At a recent productivity roundtable, federal Minister for Housing Clare O'Neil and Minister for the Environment and Water Murray Watt highlighted the need to ease the regulatory burden on builders.

They pointed to the backlog of 26,000 homes stuck in assessment under environmental protection laws as a major bottleneck. AI is expected to streamline assessments and approvals, but its precise role within the planning system remains unclear.

Will AI be limited to checking applications for completeness and classifying documents, as Victorian councils are exploring? Or will it draft written elements of assessments, as seen in the Australian Capital Territory? Could it even have some autonomy in parts of the assessment process? If so, how will it integrate with existing infrastructure, and to what extent will expert judgment be replaced?

The Temptation of a Quick Fix

Presenting AI as a quick fix for Australia's housing shortage may seem tempting, but it risks overlooking deeper systemic issues such as labor market constraints, financial and tax incentives, and the shrinking availability of social and affordable housing. The technology is also quietly transforming the planning system and the role of planners, with significant implications.

Planning involves more than paperwork; it includes judgment from site visits, stakeholder engagement, and balancing local context with broader considerations. Removing these elements can make the system and its practitioners more fragile, displacing planners’ expertise and complicating accountability when things go wrong. Moreover, errors involving AI can be difficult to trace, as explainability has been a persistent challenge for the technology.

The NSW government suggests that having a human oversee the final decision is sufficient to address these concerns. However, AI does not simply sit idle waiting for approval. It influences, frames, and shapes what gets noticed or ignored throughout the assessment process, often in subtle ways.

For instance, highlighting certain ecological risks over others can sway an assessor’s perspective, even if local communities have different priorities. When AI ranks one assessment pathway as the “best fit” based on patterns in its training data, assessors may unconsciously narrow their choices, unaware that their options have already been constrained.

Lessons from Robodebt

The Robodebt scandal offers a cautionary tale. Initially marketed as a way to improve debt recovery efficiency, the program ultimately became a $4.7 billion disaster. An automated spreadsheet—far from AI—caused harm to thousands of people, triggered a class-action lawsuit, and eroded public trust in the government.

If governments now view AI as a tool for reforming planning and assessments, they must avoid rushing in without careful consideration. The fear of missing out may be real, but the smarter approach is to pause and ask: what problem are we truly trying to solve with AI, and do we all agree it’s the real issue?

Only after addressing these questions can we tackle the harder task of implementing AI responsibly, avoiding the pitfalls that led to failures like Robodebt.

Responsible Innovation as a Path Forward

Responsible innovation means anticipating risks and unintended consequences early on by engaging with those who will use and be affected by the system. It involves proactively identifying blind spots and being responsive to impacts. There are numerous research case studies, tools, and frameworks available to guide the design, development, and deployment of AI systems in planning.

However, the key lies in addressing root causes and unintended consequences while questioning the underlying assumptions about the vision and purpose of the AI system. Ignoring the basics of responsible innovation could lead to another cautionary tale, where an AI-driven solution intended to resolve the housing crisis ends up causing similar problems as Robodebt.