Limits and Possibilities of Human-Machine Collaboration

Liu Kejun et al. (2025) study "Research on the Impact of AI Dependency by Learning Personality Type on Creativity — Focused on Graduate Students Majoring in Art/Design and Music" provides empirical data on whether AI use helps or harms creativity among creative professionals. Research design: U&I learning personality types (Action, Normative, Exploratory, Idealistic) × Guilford''s creativity components (Fluency, Originality, Flexibility, Elaboration) × ChatGPT dependency as mediating variable. Most significant finding: AI dependency has a statistically significant negative mediating effect on the creativity-performance relationship. Students with Action and Normative learning personalities had potential for direct positive creativity effects, but higher generative AI dependency significantly reduced creativity scores (negative mediation). The higher the creative capability (particularly Fluency, Originality, Elaboration), the lower the AI dependency — suggesting creatively capable students use AI as a "supplementary tool" rather than a "substitute." Critical implication: when learners over-depend on AI and skip the process of thinking independently, creativity is damaged. This is distinct from AI use causing creativity reduction — it''s AI dependency causing creativity reduction by eliminating the generative cognitive struggle that produces creative outputs. For educators and practitioners: the research suggests AI tools should be introduced with explicit frameworks for maintaining human creative agency; "AI-first" workflows that eliminate the cognitive work of generation may be systematically counterproductive for developing creative capabilities, even when they improve short-term output quality.