Trusting the Machine : Analyzing Gen Z's knowledge and usage of ChatGPT and its impact on trust in generated text

Author(s): Wilts, D.M. (2024)

Abstract:
Purpose: Ensuring the trustworthiness of chatbots like ChatGPT is increasingly important given their rapid adoption and potential societal benefits. This study focuses on Generation Z, the digital native demographic, by examining their trust, knowledge, and usage of ChatGPT. By analyzing these aspects, the study aims to identify patterns and provide practical recommendations for improving the trustworthiness of AI technologies. which is essential for fostering their optimal adoption and ensuring the responsible integration of AI technologies in society. Method: To understand trust in AI among Generation Z, 15 semi-structured interviews were conducted with individuals aged 16-28 who use ChatGPT. These interviews allowed for in-depth exploration of participants' perceptions and behaviors. Purposive sampling ensured a balanced gender representation. Result: The findings indicate that participants with a comprehensive understanding of ChatGPT's functionality exhibited significantly greater trust in its outputs. These individuals were more likely to perceive the AI's responses as reliable and accurate, as with deeper knowledge they were also more cognizant of the chatbots’ limitations. This awareness fostered a balanced trust, appreciating the AI's capabilities while acknowledging its boundaries. Participants at the younger end of the Generation Z spectrum generally discerned appropriate and inappropriate uses through trial and error. Conclusion: This study explored how Generation Z interacts with ChatGPT, revealing that deeper understanding of its capabilities enhances trust in its outputs, seen as reliable yet bounded by awareness of limitations. Younger participants learned usage norms through trial and error. Findings provide theoretical insights and practical guidance for AI scholars, developers, and educators. Future research should address limitations in sampling methods to ensure broader population representativeness.

Document(s):

Wilts_BA_BMS.pdf