Competition Shapes Students’ Use of Generative AI in Creative Problem-Solving: Behavioral Distancing and Its Individual and Collective Changes
Year: 2026
Type: article
Study design: Clinical Trial
Abstract: Generative AI can support students’ learning and problem-solving, yet little is known about whether and how human–AI collaboration changes across social contexts. This study introduces competition as an important social context and examines how it shapes students’ use of AI-generated content in creative problem-solving. A total of 92 undergraduates were randomly assigned to competition or no-competition conditions and completed three creative problem-solving tasks with GenAI support, yielding 276 student-task observations. Semantic similarity analysis, behavioral coding, regression analysis, and permutation tests were used to examine changes in AI-content use and their individual- and collective-level changes. Results showed that competition reduced students’ semantic overlap with AI-generated content and increased almost full rejection of AI suggestions, without a corresponding statistically detectable increase in self-directed improvement. Competition was associated with greater idea fluency and flexibility, but lower fluency-normalized flexibility, and neither individual semantic distinctiveness nor collective semantic diversity consistently increased. Together, these findings distinguish behavioral distancing from constructive human agency in AI-supported problem-solving: using less AI-generated content does not necessarily mean engaging with AI more independently or achieving greater relative diversity. The study therefore highlights human–AI collaboration as a socially situated process and shifts attention from the amount of AI use to the performance of students’ engagement with AI-generated content. Supporting effective AI-assisted learning may require fostering students’ capacity to critically evaluate, transform, and extend AI-generated ideas rather than simply encouraging greater or lesser reliance on GenAI.
Source: Behavioral Sciences
DOI: 10.3390/bs16112193
Institutions: Peking University, Central China Normal University
Language: English
Subfield: Experimental and Cognitive Psychology
Field: Psychology
Domain: Social Sciences
SDG: Quality education
Open Access status: gold
Estimated APC paid: $2,654
Award: G139609714
OpenAlex ID: W7236867098
