Where AI Might Actually Help Mental Health Care: Spotting Depression Earlier

A new piece in The Conversation’s New Zealand edition sets out a careful case for where artificial intelligence might genuinely assist mental health care, and where it should not go. The authors, from the University of Auckland’s 2DN research group, open with a familiar scene: a person wakes overwhelmed in the middle of the night and, instead of calling a loved one or booking a session, opens a chatbot.

Their central argument is that AI is a tool to support mental health care rather than something that can or should replace it. “An AI system may sound understanding, but it cannot truly understand the person behind the conversation,” they write. “Unlike mental health professionals, AI is not held to the same professional or regulatory standards if something goes wrong.”

The specific application the group is investigating is early detection. Depression often changes how people communicate, and the authors point to shifts in “speaking rate, pauses, tone of voice, word choice and emotional expression” as clues to a person’s mental state. Researchers call these measurable patterns “digital biomarkers”, and the Auckland work explores whether AI can learn to recognise them from both speech and text.

The framing is deliberately modest. “Rather than diagnosing people or replacing clinicians, the goal is to develop tools that support screening and monitoring, helping flag people who may benefit from further assessment,” the authors write. They compare it to how a wearable device can detect unusual heart activity without replacing a cardiologist, giving clinicians another piece of information to inform a decision.

The significance for Aotearoa is in the gap the piece names directly. The authors note that mental health services in countries such as New Zealand and Australia keep people on lengthy waiting lists, and they cite official figures on that strain. A screening aid that flags who might benefit from earlier assessment speaks to a real bottleneck, not an abstract one. Early findings suggest AI-based tools can help reduce symptoms of anxiety and depression when carefully designed and used appropriately, though the authors are clear this is conditional, not settled.

They are equally direct about the risks. Mental health data is among the most sensitive information a person can share, so privacy, security and informed consent have to be protected. AI systems can inherit biases from their training data, which may affect how well they work across different populations. And there is the problem of over-reliance: because these tools often respond in ways that feel supportive or validating, people may accept advice without questioning it or seeking professional help. The authors also warn that AI can generate inaccurate advice, sometimes reinforcing harmful beliefs and missing signs of crisis.

The conclusion draws a clean line between what technology does and what people do. “Technology can recognise patterns. People provide empathy, trust and clinical judgement,” the authors write. Their view is that the value of AI here lies in helping people understand their own wellbeing and supporting clinicians to identify risks earlier, with the two working together rather than one standing in for the other.

The full piece, including the research links behind each claim, is worth reading in the original.

Source - Read the original piece in The Conversation’s New Zealand edition: https://theconversation.com/ai-cant-replace-mental-health-therapists-but-heres-where-it-might-make-a-difference-285948