The rise of generative artificial intelligence (AIGC) is observed in many fields. The key benefit of using the technology lies in the increase in efficiency and creativity. The current research aims to explore the mechanism of human-computer collaboration creativity stimulation in generative AI used for the generation of art design. Developing a single-scale generative adversarial network (GAN) called S_OpenGAN intended for human-computer cooperation, the intelligent generation of art creation images is achieved, and the SD method is applied to evaluate the art design creation results. Based on the experimental data, it is concluded that the developed algorithm generates high-quality images with a significant diversity even when there are insufficient amounts of data. Besides, the generated results outperform the ones generated using alternative approaches such as SNI, DAT, TransEditor, and StyleGAN-2. What is more, the inference and memory consumption of S_OpenGAN is less than that of competitive approaches on all platforms, allowing generating images with the help of fewer resources. In addition, the S_OpenGAN algorithm possesses the ability to change the preferences of observers through specific changes in image elements.
