Berkeley DBT

The Berkeley DBT project is an innovative initiative that aims to enhance mental health education by making Dialectical Behavior Therapy (DBT) more accessible through the use of advanced Natural Language Processing (NLP) technologies. The project involves developing a learning-based chatbot designed to discuss DBT concepts intelligently and guide users through essential DBT skills, including mindfulness, distress tolerance, emotional regulation, and interpersonal effectiveness. This AI-driven tool is constructed to mirror a therapeutic conversation style—empathetic and thoughtful, yet clearly defined as a support tool rather than a replacement for professional therapists.

In terms of simulation and evaluation, the project focuses on crafting simulation tools that train and fine-tune the chatbot, ensuring it aligns with established DBT protocols and accurately guides users. Rigorous evaluation metrics are also implemented to assess the chatbot's precision in recommending DBT skills and its adherence to DBT standards, thus ensuring efficacy and safety in user interactions.

Through this project, we aim to address the accessibility issues prevalent in mental health resources by providing a digital tool that users can access discreetly, assisting both individuals seeking mental health support and providers. This is complemented by a strong ethical framework that prioritizes user data safety and the ethical use of AI in mental health, positioning the chatbot as an educational and supplementary resource rather than a therapeutic substitute.