Assimilating generative AI into ideation work: seeding, search and human contribution
Jafar Sabbah, Aneesh Banerjee, Feng LiPurpose
This paper examines how generative AI (GenAI) systems can be assimilated into idea generation (ideation) work so that human–GenAI collaboration augments creativity. It asks which seeding configurations with GenAI improve novelty, utility and feasibility, and examines preliminary process evidence regarding human contribution to joint search.
Design/methodology/approach
Drawing on a search-based view of creativity, we conceptualise GenAI as an information system that structures individuals' search and evaluation of information. We test four configurations with 400 professionals: human-only ideation, unstructured human–GenAI collaboration, GenAI seeding and a TC-based GenAI configuration. Expert evaluators rate novelty, utility and feasibility.
Findings
Statistically reliable improvements in novelty, utility, or feasibility were not detected for either unstructured human–GenAI collaboration or conventional GenAI seeding. The TC-based GenAI configuration, by contrast, significantly improves novelty relative to unstructured collaboration and conventional seeding. Statistically significant differences in utility were not detected relative to either GenAI comparison condition, while feasibility was lower relative to conventional seeding.
Practical implications
Organisations should not assume that simply enabling GenAI in ideation tools will improve creative outcomes. When seeking to promote novelty, organisations may consider the full TC-based GenAI configuration tested here, combining a short TC orientation with TC-based GenAI-generated seeds.
Originality/value
The study advances information systems research on AI assimilation and human–AI collaboration by showing that creative augmentation is not automatic and depends on how GenAI is configured to structure joint search. It identifies the TC-based GenAI configuration as a concrete way to enhance novelty in human–GenAI ideation.