Massachusetts Institute of Technology

MEASURING COLLECTIVE INTELLIGENCE

Thomas W. Malone, Stephen Kosslyn (Harvard), Anita Williams Woolley (Carnegie Mellon), Christopher Chabris (Union College) and David Karger

Intelligence tests can predict the performance of individuals across a broad range of tasks. Imagine if we had an instrument that could predict the performance of groups—combinations of people assisted by computers, telecommunications links, and other man-made devices—across a range of relevant tasks. And imagine that this instrument would allow us to test whether efforts to improve performance on key tasks actually succeeded in making a group “smarter.”

The goal of the Measuring Collective Intelligence project is to find out whether such an instrument is feasible, and if so, to develop and test it, and then to use it to assess the effectiveness of interventions designed to enhance performance. The project will focus in three main areas.

  • First, we will use what is already known about measuring individual intelligence to generate, by analogy, ideas about new ways of measuring the collective intelligence of groups. We will determine whether the striking pattern of correlation in individuals’ performance across a wide range of tasks even exists for human-machine groups. Then we will develop statistically validated tests for measuring the key components of collective intelligence in human-machine groups.

  • Next, in order to better understand the “active ingredients” of collective intelligence, we will combine what is already known about how groups of people interact effectively with new approaches for modeling the information processing that occurs in human-machine groups. A key goal will be to find critical factors (such as group size, communication patterns, or individual capabilities) that determine a human-machine group’s performance across a wide range of tasks.

  • Finally, we will use the results of these first two approaches to develop and test computational tools or other methods for increasing the collective intelligence of groups. For instance, we may develop and test new tools for filtering information more efficiently or for combining the judgments of multiple people more effectively.

*Harvard University Department of Psychology

 

If you are interested in participating in this study, please click here.