From Search Engines to Generative AI: Rethinking Human Creativity in the Digital Age

Abstract

Digital tools promise to extend human creativity, but do they deliver? Across four between-subjects experiments, participants generated creative gift ideas with assistance from web search, a large language model (LLM), or no digital support. Working with an LLM produced the most creative ideas, whilst the theoretical advantages of web search did not materialise in practice - differences between web search and the control group were inconsistent. Crucially, the creativity boost from LLM use was moderated by experience: participants more familiar with AI generated more creative ideas, suggesting these gains are not equally accessible. Analysis of think-aloud protocols, and, for LLM users the creative processes embedded in their prompts, revealed that the creative process operates differently across the three conditions, with distinct phases of the creative process contributing differently to idea quality. Elaborating on an initial idea was instrumental to creativity across all groups, whilst problem construction and evaluation had a lesser - and sometimes negative - impact. Whatever digital means we employ to enhance everyday creativity, it is the human creative process that ultimately underpins their effective use.

Date of Award	17 Jul 2026
Original language	English
Awarding Institution	University of Winchester
Supervisor	Paul Sowden (Supervisor)