With a growing interest in Generative AI, this study investigates the impact of different prompting approaches, including base, zero-shot, and few-shot learning, and the role of human intervention in generating creative design solutions using GPT-4. Specifically, it distinguishes exemplary ideas that are superior in terms of creativity from those that are not, and investigates how different inputs in training influence the resulting creative performance. The study results revealed that zero-shot prompts, especially those emphasizing diversity, produced the most varied solutions. Few-shot prompts, using high-quality (top) and low-quality (bottom) examples, demonstrated that top examples resulted in greater diversity compared to bottom examples. The findings highlight the crucial role of human judgment in input training, noting that high-quality examples broaden the space in few-shot learning. Therefore, this research suggests insights into the necessity of integrating Gen-AI with human intervention by demonstrating how human-AI collaboration can be fine-tuned to maximize benefits in creative endeavors.