The mental health field has spent two years arguing about chatbots. A group of researchers led by the University of Hong Kong wants the conversation to widen before the tools harden into infrastructure.
Writing in Psychotherapy and Psychosomatics, the team set out a roadmap for AI and adjacent technologies in mental health care, drawn from collaborators at Cambridge, UC Berkeley, the National University of Singapore and other centres. Professor Benjamin Becker of HKU’s Department of Psychology led the work.
The central proposal is personalised care assembled from several signals rather than conversation alone. Brain measures, clinical symptoms, life history and social circumstances would combine to select and adapt treatment for individual patients.
The authors also look past chatbots to brain imaging, smartphone-based monitoring and AI-informed brain stimulation, framing these as tools for assessment, treatment and follow-up. They are candid that all of it is early, and that evaluation must join neuroscience, psychology, clinical expertise and patient perspective.
The sharpest line is a limit. Conversational tools must be grounded in evidence-based therapy to help with assessment, engagement and symptoms, and current evidence does not support handing high-risk management to an autonomous system or replacing human therapy.
That position lands as US states write rules for AI therapy and health systems experiment with triage bots. The roadmap’s value is timing: it arrives while the design space is still open.
