Usability of KAI: A generative AI chatbot to support mental health among university students in Latin America

Abstract

Mental health among university students in Latin America has been compromised by stress, depression, and anxiety, which have been exacerbated by the pandemic. Despite the vulnerability of this population and the potential of digital tools, the actual adoption of chatbots for psychological support remains low due to doubts about their effectiveness and ease of use. The purpose of this study is to analyze the usability of KAI, a virtual assistant powered by generative artificial intelligence designed to assess symptoms using the DASS-21 scale and offer personalized recommendations. The methodology employed a mixed-methods descriptive and exploratory approach with 112 students from three Latin American universities, using the System Usability Scale (SUS) and qualitative interviews with psychology students. The main results indicate that the chatbot has outstanding usability and high acceptability, with a positive perception of its ease of use and security, although significant differences in evaluation were identified among the participating institutions. The main conclusion highlights that KAI is an intuitive and technologically viable tool that functions as a scalable emotional support resource, provided it maintains strict professional referral protocols in cases of high severity.