Navigation auf uzh.ch

Suche

Institute of Biomedical Ethics and History of Medicine (IBME)

New Publication: «The Role of Humanization and Robustness of Large Language Models in Conversational Artificial Intelligence for Individuals With Depression: A Critical Analysis»

We are pleased to announce the publication of a new article by Andrea Ferrario, Jana Sedlakova and Manuel Trachsel, MD, PhD in JMIR Mental Health.

Title: "The Role of Humanization and Robustness of Large Language Models in Conversational Artificial Intelligence for Individuals With Depression: A Critical Analysis"

Summary: In this paper, the authors address two critical challenges in the use of large language model (LLM)-based conversational AI for mental health applications: the unwarranted humanization of these systems and their lack of contextualized robustness. Drawing from philosophy, psychiatry, and computer science, the authors elaborate on these issues and propose solutions to mitigate the risks faced by vulnerable users, such as individuals with depression. The authors also caution against using terms like "being empathetic" when referring to LLM-based conversational AI.

Read the full paper.