关键词: 人智协同, 人智协同顺序, 创造力, 锚定效应, 心理所有权

Abstract: As organizations increasingly integrate artificial intelligence (AI) into creative work, a critical managerial question arises: how should human-AI collaboration be structured to maximize creativity? Prior research has generally favored a human-first collaboration sequence. However, given the multi-stage nature of the creative process, it remains unclear whether this advantage generalizes across different stages. Drawing on idea journey theory and the anchoring effect, we argue that a human-first collaboration sequence in the idea elaboration phase heightens psychological ownership, leading creators to resist subsequent AI-generated structural suggestions. This resistance hinders improvements in idea articulateness, thereby constraining overall creativity. We further propose that trust in AI moderates this process by shaping individuals’ receptiveness to AI input. We tested these arguments across three complementary experiments (N = 741) involving diverse creative tasks and samples. Study 1 recruited 230 full-time employees to complete a product design elaboration task. Study 2 was a field experiment involving 271 participants developing marketing plans for newly launched coffee products, including a no-AI control condition. Study 3 recruited 240 full-time employees to extend these findings by identifying psychological ownership as a key mechanism and by testing trust in AI as a boundary condition that mitigates the disadvantage of a human-first sequence. Across the three studies, we find consistent support for our hypotheses. In idea elaboration tasks, a human-first collaboration sequence can be detrimental to creativity. Specifically, compared to an AI-first collaboration sequence, a human-first sequence is more likely to induce stronger psychological ownership of initial ideas, which in turn inhibits improvements in idea articulateness and ultimately reduces overall creativity. Further analyses show that trust in AI moderates this process. Under a human-first sequence, higher levels of trust in AI alleviate cognitive anchoring and increase individuals’ willingness to incorporate AI-generated suggestions, thereby improving the effectiveness of idea elaboration. This research makes three contributions to the literatures on human-AI collaboration and creativity. First, it challenges the assumption that a human-first sequence is universally advantageous by demonstrating that its effectiveness is contingent on the creative phase. Second, it uncovers a cognitive mechanism explaining why collaboration may remain suboptimal even when AI is used: a human-first sequence strengthens self-anchoring and psychological ownership, prompting creators to discount or overlook AI’s structural input. Third, by shifting attention from whether to use AI to how and when to use it, this research provides new insights into how organizations can better harness the joint potential of humans and AI in creative problem solving.

Key words: Human-AI collaboration, Human-AI collaboration sequence, Creativity, Anchoring effect, Psychological ownership