AI-Powered Chatbot Pathway for Supporting Gender-based Violence Survivors in Turkiye

This thesis investigates the adaptation of best-fit chatbot technology in Turkish context that AI- driven chatbots are one of the fastest advancing technologies today. The specific three research questions aimed to find the following answers: key elements/components of recognized AI chatbot technologies being used by reputable agencies and organizations to support GBV survivors and programming, adapting possible features to the Turkish context, and uncovering the best AI-driven chatbot solution. To reveal the potential and challenges of chatbot to the Turkish context following the earthquakes centered in Kahramanmaraş on February 6, 2023, both primary and secondary data were collected. In the analysis of secondary data, 14 chatbots designed and launched supporting to violence survivors worldwide were examined, and a total of 6 different features were identified by listing all the features of these chatbots: risk assessment, case management, psychological support, legal counseling, service advisor, and staff support. In the second stage of the study, in-depth interviews were conducted with 13 humanitarian workers who work in various organizations such as UN, INGOs, NGOs, and donor organization, especially in the field before and after the earthquake, regarding the chatbot to be adapted to the post-earthquake context. Key informant interviews were analyzed using the ATLAS-TI software. Except for one participant, all participants expressed that AI is an important part of their daily lives, but full adaptation has not yet been achieved in the organizations they work for, with 9 out of 13 participants stating that AI is not used in projects and programs at their workplace. The participants evaluated the positive aspects (Positive Coding: 51 times) of AI more than its negative aspects (Negative Coding: 38 times). Considering the needs, demands, and support in the field, the features considered most important for the chatbot designed to provide support to GBV survivors were service advisor and legal counseling. The psychological support feature was considered the least adaptable feature both technically and in terms of preference by the affected people. Additionally, concerns and contributions raised by the participants regarding issues such as data privacy, algorithmic bias, and misinformation were compiled and included.