Can AI Help in Crowdsourcing? A Theory-Based Model for Idea Screening in Crowdsourcing Contests
(A2023-113945)

AUTHORS

Jason Bell, University of Oxford; Christian Pescher, Universidad de los Andes; Gerard Tellis, University of Southern California; Johann Füller, University of Innsbruck

ABSTRACT

Crowdsourcing generates thousands of ideas. The selection of best ideas is costly because of the limited number, expertise, objectivity, and attention of judges. The authors test two theoretical models and seven model specifications for idea selection. Results are the following. First, current models are unable to replace humans in selecting the best ideas. Second, however, these models do an excellent job in screening bad ideas, reducing judges’ tedium and enabling their focus on the best ideas. Third, the authors develop an Idea Screening Efficiency curve that trades off the False Negative Rate of good ideas screened out with the rate of ideas screened. Managers can choose the desired point on this curve for optimal idea screening. Fourth, a new predictor, Word Atypicality, is simple and efficient in such screening. Theoretically, this predictor selects ideas that are detailed, rich, and inclusive.