HOW AI-SUPPORTED SEARCHES THROUGH OTHER PERSPECTIVES AFFECT IDEATION OUTCOMES
Julian Wahl, Katja Hutter, Johann Füller
Abstract
Seeking inspiration from other perspectives is a prominent mechanism to support ideation. AI-based language models can help overcome information processing limits and efficiently structure large solution spaces spanned by prior ideas. However, it remains unclear how the search through a solution space affects the subsequent idea generation. This study explores the influence of different sets of prior idea stimuli pre-structured by an AI-supported clustering on ideation outcomes. The sets varied in quantity and semantic diversity. In a survey experiment, 181 participants generated 447 ideas evaluated according to major idea performance characteristics. Results indicate that seeing an extensive set of ideas from various clusters improves idea novelty and positively and semantic diversity. In a survey experiment, 181 participants generated 447 ideas evaluated according to major idea performance characteristics. Results indicate that seeing an extensive set of ideas from various clusters improves idea novelty and positively interacts with domain-specific knowledge. However, it negatively affects idea feasibility and specificity. These findings encourage innovators seeking particularly novel ideas to complement their current processes with AI-supported clustering tools while taking steps to avoid vagueness.