The Effectiveness of LLMs in Mental Health

Abstract: It has an impact on everyone's health, which is why mental illness should be prioritized in the healthcare industry. However, it appears that this field is developing at a somewhat slow pace. AI (Artificial-Intelligence) technologies have recently received a lot of attention in a variety of fields, including mental health. Advanced AI approaches and machine learning algorithms have made it possible to provide personalized care that primarily focuses on providing emotional support tailored to a specific individual. We explore the possibility of using large language models like OpenAI’s GPT3 and Facebook’s Llama and Stanford’s Alpaca to provide an effective conversational partner to people suffering with such mental health conditions where it may be helpful, such as depression and anxiety disorders. We compare the performance of the chatbots based on their responses to questions from counselchat.com dataset of therapist responses, and use the GPT4-davinci, the largest GPT4 model, as a judge to evaluate the quality of responses.

Keywords: AI (Artificial intelligence), LLMs (Large Language Models), Fine-tuning, chatbot, depression, anxiety, mental health.