Estudo exploratório sobre o uso do ChatGPT como apoio no design da interação humano-computador

Abstract

The advancement of Large Language Models (LLMs) has driven new forms of support in various fields, including Human-Computer Interaction (HCI). However, the adoption of these tools still requires studies to assess their impact on HCI practices. This study investigates the potential of these models, particularly the ChatGPT tool, in supporting user interface design processes, analyzing how they can contribute to idea and artifact generation, concept refinement, and decision-making. Using an exploratory methodological approach applied to a real use case, the study seeks to understand the strengths and limitations of generative AI as support for techniques commonly used in HCI. Four techniques were selected: Questionnaires, Interviews, Personas, and Scenarios. The results generated by ChatGPT were evaluated by four experts with a Ph.D. degree and experience in teaching HCI. The findings suggest that while ChatGPT can enhance productivity and produce coherent and structured results, challenges remain such as a lack of deep contextual understanding. It is concluded that the effective integration of LLMs into interface design requires approaches that mitigate these limitations and promote a critical and conscious use of these technologies, maximizing their applicability without compromising the quality and autonomy of creative processes.