How the Ideation Process Shapes the Creative Output in Innovation Contests—An Analysis Using a Large Language Model

Abstract

Innovation contests represent a type of open innovation that companies can use for gaining access to innovation ideas from different target groups. These innovation ideas have a measurable creative output and arise in an ideation process that usually takes place unconsciously among the participants, but can be traced back through operationalization with the inventive principles from the Theory of Inventive Problem Solving (TRIZ). We evaluate these two aspects—creative output and ideation process—on the basis of over 600 innovation ideas collected from more than 1200 students during eight identical courses on ‘Innovation Management’ between 2015 and 2022, using a Large Language Model (GPT-4). We identify two important relationships between the two variables. Firstly, our data show that the more inventive principles are used in the ideation process, the higher the creative output of an innovation idea. Secondly, at least in our setting, universal inventive principles outperform technological inventive principles in terms of creative output. This study is the first to use GPT-4 to simultaneously assess both the creative output of more than 600 innovation ideas and the hidden TRIZ inventive principles behind them.