Mental Health

What the evidence says

The relationship between job loss and mental health is one of the better-established findings in social epidemiology. Unemployment is associated with significantly elevated rates of depression, anxiety, and suicide — effects that persist even after controlling for pre-existing conditions and income reduction. The psychological harm of job loss is not simply poverty: it includes loss of routine, loss of social connection, loss of identity and purpose, and loss of the sense that effort leads to reward.

For workers displaced by AI, these effects may be compounded by a specific form of humiliation: the sense that your skills, accumulated over a career, have been made worthless not by your own failure but by a technological change you had no part in shaping. Research on "automation anxiety" finds elevated distress even among workers who haven't yet been displaced — the threat alone has measurable effects on wellbeing.

NZ enters this challenge with a mental health system already at capacity. Wait times for public mental health services are long; workforce shortages in psychiatry, psychology, and counselling are severe; Māori and Pacific communities face additional barriers to access.

Options for policy response

Embedded mental health support in transition programmes: Rather than treating mental health as a separate service that displaced workers must seek out, embed it as a standard component of all income support and retraining programmes. This could be as simple as mandatory initial wellbeing check-ins, or as comprehensive as co-located counsellors in employment services. Who favours this: those who believe prevention is cheaper than treatment, and that vulnerable people don't seek help unless it's offered. The debate: mandatory vs optional — mandatory reaches more people but may not be appropriate in all contexts, and raises questions about clinical autonomy.

AI therapy tools: Digital mental health tools using AI — apps like Woebot, Wysa, and others — have shown promise in randomised controlled trials for mild to moderate depression and anxiety. They are available at scale, asynchronously, and at low cost. For a stretched system, they could extend reach significantly. Who favours this: those focused on access and scale. The risk: AI therapy tools are not appropriate for severe presentations, and there is a concern that they become a cost-cutting substitute for human care rather than a complement to it. There are also specific risks around crisis situations — AI tools that fail to recognise suicidality appropriately have caused harm. Regulatory oversight is minimal in NZ at present.

Community resilience approaches: Mental health outcomes track closely with social connection and community cohesion. Programmes that invest in community infrastructure — sports clubs, volunteer organisations, local arts, community hubs — provide protective effects that clinical services cannot replicate. In regions facing significant displacement, community resilience investment may be as important as clinical capacity. Who favours this: those who believe that mental health is fundamentally a social problem, not an individual one. The risk: harder to measure, harder to fund through conventional health budgets, and may feel inadequate when people are in acute distress.

The purpose question

Perhaps the deepest challenge is one that income and clinical support cannot fully address: the question of meaning. For many people, work is not just income — it's identity, structure, contribution, and social belonging. Policies that maintain income while leaving people without meaningful activity are not fully addressing the problem.

This raises questions that go beyond the welfare system: about the value society places on unpaid care work, creative work, and community work; about whether shorter working weeks could share available employment more broadly; about education systems that cultivate intrinsic motivation rather than purely vocational skills.

These are not easy policy questions. But any mental health strategy for the AI transition that doesn't grapple with them is treating the symptom rather than the cause.