Quantifying Collective Intelligence in Human Groups

PNAS May 25, 2021 118 (21) e2005737118; https://doi.org/10.1073/pnas.2005737118

Commentary

Collective intelligence (CI) can be important for solving many scientific, business, and other problems, but groups may not be able to achieve this.

In this study, we analyzed data on group performance from 22 studies involving 5,279 individuals in 1,356 groups. The results support the conclusion that a robust CI factor characterizes the ability of groups to collaborate across task sets. We further show that CI predicts performance on an out-of-sample criterion task, mediated by the average social perceptions of group members and predicted by the proportion of women in the group.

Overall, the group collaboration process is shown to be more important in predicting CI than the skills of individual members.