Design Requirements for Responsible AI Tools in Education: Empirical Study on ChatGPT and Creative Thinking

Abstract

The integration of generative artificial intelligence (AI) tools into educational contexts presents significant opportunities as well as ethical and cognitive challenges, particularly regarding their impact on students' creative and independent thinking. While generative models such as ChatGPT are increasingly employed to support learning and ideation, their implications for human creativity and responsible AI design in requirements engineering (RE) remain underexplored. This empirical study uses educational settings as a model for RE practice, investigating both perceived and actual effects of ChatGPT on core dimensions of creative thinking—fluency, originality, and flexibility—among undergraduate students. Through a structured survey (N=100) and a controlled experiment (N=60), we found that ChatGPT significantly enhances idea fluency but does not affect originality or flexibility. The study also identifies perceived trade-offs between creative productivity and cognitive autonomy. These findings inform a set of responsible design requirements for AI systems from an RE perspective, offering guidance for the implementation of educational and professional tools that support, rather than supplant, human creativity in requirements processes.

Ähnliche Arbeiten