GenAI and mental health – Elena Walsh – University of Wollongong

Human beings are social and dependent creatures. We rely on friends, romantic partners, family, communities, therapists, and other confidantes for support, insight, and understanding. And yet, we have recently entered an era in which many now seek support from artificial agents powered by generative AI. These AI agents are increasingly used — by design or request — to simulate roles we once thought only human beings could play. Among the most rapidly growing applications is the use of large language models (LLMs) to provide emotional support or to simulate certain types of therapeutic dialogue. The talk first places this development in context of a brief characterisation of psychotherapy as a ‘living tradition’ in the sense of MacIntyre (1981): not a rigid and fixed set of practices, but a set of goals and methods that are continually critiqued and reinvented over time. Two aspects of therapeutic dialogue are singled out for comparison against LLM-based emotional or therapeutic dialogue. The first is the role of empathy in treatment. The second is the capacity of dialogue to restore ‘hermeneutical justice’ (Fricker, 2007) — that is, the restoration of vocabularies that allow experience to be accurately named and understood. The dimensions of empathy and hermeneutical justice are used as a framework to compare traditional human-to-human therapy against LLM-based dialogue or support. The talk concludes by linking the rise of LLM use for emotional or therapeutic support to globally under-resourced mental health care systems and significant barries to accessing mental health care, especially for vulnerable populations.

Bio: Elena Walsh works across the Philosophy of Psychology, the Philosophy of Science, and the Philosophy of AI. She has expertise in the study of emotion and emotional dispositions, drawing especially on dynamical systems theory, life history theory, and predictive processing models of mind. Her current research places contemporary research on emotion in dialogue with the rapidly-developing approaches to machine learning coming to define 21st-century notions of both artificial and biological intelligence. She is interested in how norms and values may be embedded into decision-making processes undertaken by AI and data-driven technologies, and how human interaction with new technologies can impact our characters and regulate our attentional and emotional capacities.

She has expertise in related areas including Moral Psychology (especially the relationship between emotion and reason) and Epistemology. She has a longstanding interest in Buddhist, Asian and comparative approaches to philosophy. Her other philosophical interests include the role of emotion and motivation in intelligent systems, and the opacity and ethical governance of emerging AI. Elena completed her PhD in 2019 at the University of Sydney. Her dissertation adopted a broadly naturalistic approach to provide a theoretical framework that explains how emotional dispositions are constructed in individuals over time.

She has previously worked for the Department of Premier and Cabinet as a policy advisor, and as a researcher at the Practical Justice Initiative at the University of New South Wales.