Results. Human vs GPT.csv

This study investigates the creative capabilities of ChatGPT compared to human students in the context of divergent thinking tasks conducted in the Ukrainian language, a novel linguistic framework for such comparisons. By analyzing responses to the Unusual Uses Test, where participants generated creative uses for a stool, we evaluated fluency, flexibility, and originality of ideas from 43 students (aged 12–15) and ChatGPT 3.5 across 28 prompts. Results indicate that ChatGPT significantly outperformed students in fluency, but students surpassed ChatGPT in relative flexibility, demonstrating greater diversity in their ideas. While no significant differences were found in originality, human responses exhibited greater variability, including highly unique ideas absent in ChatGPT’s outputs. ChatGPT’s responses, though detailed, were more conventional and contained linguistic inaccuracies in Ukrainian, likely due to limited training data.