Abstract: Focusing on the evaluation of user creativity in human-AI collaboration contexts, this study systematically investigates the effects of AI collaboration and initial creativity levels on users' creative thinking performance. A behavioral experiment was conducted, where 95 participants were recruited and divided into experimental and control groups. The Williams Creativity Aptitude Scale was employed to assess their initial creativity levels, based on which participants were categorized into high and low creativity groups. Subsequently, within the four task scenarios of the Torrance Tests of Creative Thinking, comparative analyses were performed on participants' behavioral characteristics and creative performance in both verbal and pictorial creation tasks. The findings indicate that AI collaboration exerts a positive enabling effect on users' overall creative performance, though this effect is moderated by task characteristics. Additionally, initial creativity level serves as the core foundation of creative performance and exhibits a synergistic effect with AI collaboration. The research has revealed the positive promoting effect of AI on creativity performance, and further explored the differentiated characteristics of users with different levels of creativity in human-computer collaboration, providing empirical evidence for the rational utilization of AI and the sustainable development of human-computer collaboration.

Key words: Artificial intelligence, Human-AI collaboration, Creativity, Information behavior, Creative thinking