“Responsible AI for Mental Health” Workshop: Outputs and Insights

We were pleased to have ~65 in-person attendees and more than 125 online participants from around the world. The event featured a diverse lineup of 15 international keynote speakers and panellists across two thematic sessions: “Digital Mental Health” and “Responsible Use of AI in Mental Health”. On the next day, September 13th, we continued with an invitation-only workshop, where experts discussed, through structured design-thinking activities, preliminary policy recommendations for the responsible application of AI in Mental Health. You can watch the recordings of the event on our YouTube channel and read the White Paper documenting our findings and preliminary recommendations:

Morning Session: Digital Mental Health

To conclude the session, Prof. Elvira Pérez-Vallejos from the University of Nottingham spoke about the application of digital tools for death and aging, highlighting important insights in this often-overlooked aspect of mental health care.ing our findings and preliminary recommendations:

Afternoon Session: Responsible Use of AI in Mental Health

Prof. Maria Liakata from Queen Mary University of London then spoke about the socio-technical limitations of large language models (LLMs) for social computing. The day ended with a panel featuring Dr. Laura Ihle (Northeastern University)Dr. Joseph Connor (CarefulAI)Mariana Silva (University of Cambridge), and Dr. Brieuc Lehmann (University College London), moderated by Dr. Stuart E. Middleton from the University of Southampton. They discussed the challenges in applying AI responsibly in the mental health sector, including ethical considerations and data-driven tools. Co-organiser Dr. Annika Marie Schoene closed the event with some thoughtful remarks.

Participant Feedback

Gathering participant feedback and encouraging discussions were essential aspects of the event. We collected feedback through multiple methods, including Q&A sessions for both in-person and online attendees, in-person surveys using stickers on posters, and online polls. These methods allowed participants to help shape our next steps towards the responsible application of AI in mental health by sharing their own policy recommendations, as well as the risks, opportunities, and challenges they envisioned. Participants highlighted the need for more funding in mental health care, ethical frameworks, and market regulations, among other key points.

Feedback from participants emphasised the importance of collaboration among diverse sectors, ensuring ethical development, and implementing transparent AI practices. Many attendees noted the value of multidisciplinary dialogue and expressed interest in ongoing opportunities to engage in responsible AI practices for mental health.

Follow-up Event

Using user-centered design principles, participants worked in groups using inputs from surveys collected the day before as a starting point. They engaged in four activities: grid analysisprioritisationscenario modelling, and drafting preliminary recommendations. These recommendations will be shared with the broader community soon in an official report.