Mental Health - Large Language Models and the Impending Generative AI Tsunami: Whatever Shall We Do?
Tracks
Diamond Ballroom III: In-Person Only
| Thursday, June 5, 2025 |
| 2:30 PM - 2:50 PM |
Overview
Andrew J Greenshaw PhD, University of Alberta & Taimi Allan, Mental Health Commissioner for South Australia
Presenter
Taimi Allan
Mental Health Commissioner for South Australia
Mental Health - Large Language Models and the Impending Generative AI Tsunami: Whatever Shall We Do?
Abstract
Large language models (LLMs) like GPT-4 are the drivers of generative AI apps – chatbots, information navigators, internet app designers, assistive AI tools of many kinds. They are at the forefront of offering digital health solutions but offer both opportunities and risks. Taimi Allen and Andy Greenshaw will provide a dialogue to inform conference attendees about what large language models (LLMs) are and their range of application in providing potential digital mental health solutions. The risks and opportunities of using these applications in the public domain will be discussed, with examples, and reduction of “risks” in this context will be considered in terms of putting limits on sources of information for training LLMs and putting ‘guardrails’ around the types of responses permitted in “conversational” applications. The importance of co-creation and the voices of those with lived and living experience of mental health challenges will be emphasized. The presentation will help conference attendees to understand the potential of these LLM applications for supporting mental health in the community. We hope to provide a glimpse of the digital therapy assistant of the future and illustrate how anyone can customize a LLM app for personal use.
Three Key Learnings:
1. Be informed about what large language models (LLMs)are and what is their range of application in digital mental health.
2. Be informed about the risks and opportunities of using these applications in the public domain and understand how risks can be reduced by putting limits on sources of information for training LLMs and putting ‘guardrails’ around the types of responses permitted in “conversational” applications.
3. Understand the current and likely future state of these LLM applications for supporting mental health in the community. Catch a glimpse of the digital therapy assistant of the future and learn how anyone can customize a LLM app for personal use.
Three Key Learnings:
1. Be informed about what large language models (LLMs)are and what is their range of application in digital mental health.
2. Be informed about the risks and opportunities of using these applications in the public domain and understand how risks can be reduced by putting limits on sources of information for training LLMs and putting ‘guardrails’ around the types of responses permitted in “conversational” applications.
3. Understand the current and likely future state of these LLM applications for supporting mental health in the community. Catch a glimpse of the digital therapy assistant of the future and learn how anyone can customize a LLM app for personal use.
Biography
Taimi Allan is the Mental Health Commissioner of South Australia, appointed in October, 2023. With a 20-year career in mental health, Taimi is a passionate advocate for incorporating lived experience into mental health care. Previously serving as Director of Ember Innovations in New Zealand, where Taimi implemented innovative solutions to mental health system gaps, Taimi’s experience includes leading mental health consumer advocacy organizations, co-designing accessible wellbeing programs, and producing empathy-building virtual reality films. Taimi's dedication to improving mental health through creativity, technology, and face-to-face connection has earned her numerous national and international accolades.
Dr Andrew J. Greenshaw PhD
Professor of Psychiatry and Neuroscience
University of Alberta, Canada
Mental Health - Large Language Models and the Impending Generative AI Tsunami: Whatever Shall we Do?
Abstract
Large language models (LLMs) like GPT-4 are the drivers of generative AI apps – chatbots, information navigators, internet app designers, assistive AI tools of many kinds. They are at the forefront of offering digital health solutions but offer both opportunities and risks. Taimi Allen and Andy Greenshaw will provide a dialogue to inform conference attendees about what large language models (LLMs) are and their range of application in providing potential digital mental health solutions. The risks and opportunities of using these applications in the public domain will be discussed, with examples, and reduction of “risks” in this context will be considered in terms of putting limits on sources of information for training LLMs and putting ‘guardrails’ around the types of responses permitted in “conversational” applications. The importance of co-creation and the voices of those with lived and living experience of mental health challenges will be emphasized. The presentation will help conference attendees to understand the potential of these LLM applications for supporting mental health in the community. We hope to provide a glimpse of the digital therapy assistant of the future and illustrate how anyone can customize a LLM app for personal use.
Three Key Learnings:
1. Be informed about what large language models (LLMs)are and what is their range of application in digital mental health.
2. Be informed about the risks and opportunities of using these applications in the public domain and understand how risks can be reduced by putting limits on sources of information for training LLMs and putting ‘guardrails’ around the types of responses permitted in “conversational” applications.
3. Understand the current and likely future state of these LLM applications for supporting mental health in the community. Catch a glimpse of the digital therapy assistant of the future and learn how anyone can customize a LLM app for personal use.
Three Key Learnings:
1. Be informed about what large language models (LLMs)are and what is their range of application in digital mental health.
2. Be informed about the risks and opportunities of using these applications in the public domain and understand how risks can be reduced by putting limits on sources of information for training LLMs and putting ‘guardrails’ around the types of responses permitted in “conversational” applications.
3. Understand the current and likely future state of these LLM applications for supporting mental health in the community. Catch a glimpse of the digital therapy assistant of the future and learn how anyone can customize a LLM app for personal use.
Biography
Andy Greenshaw is a Professor of Psychiatry and Neuroscience at the University of Alberta. Training in Europe and Canada before joining the University of Alberta in 1986, Dr. Greenshaw has held various leadership roles, including Associate Vice President (Research) and Scientific Director of the APEC Digital Hub for Best Practices in Mental Health. He is also a Fellow of the Royal Society of Arts and has contributed extensively to mental health research, focusing on biological psychiatry and behavioral neuroscience. Andy’s work includes applying machine learning to predict mental health treatment outcomes
