Building safe AI: how we train our models for mental health conversations

Join Ricardo Rei, Head of AI Research and Maya, Head of Clinical, for a critical discussion on AI Safety in Mental Health. In this webinar, we will explore the significant challenges that LLMs introduce into the sensitive domain of mental health, including the profound difficulty in accurately evaluating their performance in clinical scenarios. We will then examine the crucial role of guardrails, making the case that generic safety measures are insufficient and highlighting the necessity of building specialized guardrails tailored for mental health. Finally, we will outline a forward-looking approach to overcoming these challenges by co-designing new foundational models from the ground up, working in close collaboration with clinical teams to ensure safety and efficacy are built in by design.