Scaffolding Creativity: How Divergent and Convergent LLM Personas Shape Human Machine Creative Problem-Solving

TL;DR

This paper investigates how two explicit LLM personas—divergent and convergent—shape human creative problem solving in a co-creative setting. A randomized experiment compares persona-guided interaction against a standard LLM, revealing that user perceptions of creativity and engagement depend on trait-based preferences and mode switching. Computational creativity metrics show increased originality under persona guidance without inflating idea quantity, while authorship shifts toward collaborative processes. The findings offer design principles for creativity support systems that balance exploration and convergence through persona-based guidance and personalization, advancing human-AI collaboration that scaffolds rather than substitutes human creativity.

Abstract

Large language models (LLMs) are increasingly shaping creative work and problem-solving; however, prior research suggests that they may diminish unassisted creativity. To address this tension, a coach-like LLM environment was developed that embodies divergent and convergent thinking personas as two complementary processes. Effectiveness and user behavior were assessed through a controlled experiment in which participants interacted with either persona, while a control group engaged with a standard LLM providing direct answers. Notably, users' perceptions of which persona best supported their creativity often diverged from objective performance measures. Trait-based analyses revealed that individual differences predict when people utilize divergent versus convergent personas, suggesting opportunities for adaptive sequencing. Furthermore, interaction patterns reflected the design thinking model, demonstrating how persona-guided support shapes creative problem-solving. Our findings provide design principles for creativity support systems that strike a balance between exploration and convergence through persona-based guidance and personalization. These insights advance human-AI collaboration tools that scaffold rather than overshadow human creativity.

Figures (6)

• Figure 1: Persona-Guided Chat Interface used in the experiment.
• Figure 2: Post-session questionnaire outcomes by experimental condition.
• Figure 3: Distribution of persona preference ratings (1-4 scale).
• Figure 4: Significant personality trait-outcome correlations in the treatment condition. Error bars represent 95% confidence intervals around Pearson's r.
• Figure 5: Question mark frequency in mid-to-late conversation phases.