Artificial Social Intelligence for Successful Teams (ASIST) Study 4 Dragon Testbed Dataset was developed in a human subjects research study designed to assess the capability of artificial intelligence to instantiate a Machine Theory of Teams and apply it to generate and issue (or withhold) advice to team members that improve team state (e.g., motivation), process (e.g., synchronization), and mission effects (e.g., game score). These agents -- called Artificial Social Intelligence Advisors (ASI Advisors) -- draw measurements of team states and processes from agents called Analytic Components (AC). They take inputs from survey responses and behaviors of a three-person team executing a bomb disposal task in Minecraft.
This material is based upon work supported by the Defense Advanced Research Projects Agency (DARPA) under Contract No. HR001119C0130. Any opinions, findings conclusions, or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the DARPA.
Data Overview
The dataset was collected between 2023-08-03 and 2023-11-20. This dataset consists of 1160 games that had valid end states, and 1112 of them included post-trial surveys. Each game has one Zip file (e.g., 230804001812+P000003_P000004_P000005_1+NO_ADVISOR+9c5f329e-0c52-4254-b165-60f3a57b4fd3.zip). The Zip file name starts with a UTC in the form of YYMMDDHHMMSS, followed by the team name in the format of the three participants' ID in ascending order and the number of games the team has played together, followed by one of three advisor types (i.e., NO_ADVISOR, ASI_DOLL_TA1_RITA, or ASI_CMU_TA1_ATLAS), and followed by the unique 36 digit trial ID. Inside the Zip file, it has seven files: the testbed data in .metadata format, an overview of the testbed data in JSON format, and five CSV files (i.e., agent_tests, chat_measures, individual_measures, intervention_measures, trial_measures). The metadata files include all the raw data, such as surveys, chat messages, state of the task environment, etc. The CSV files are data extracted from the metadata files to show certain aspects of the data variables for the convenience of those not used to reading metadata files.
The Study 4 dataset has a size of 4GB.
A readme file (README.txt) describes the dataset contents in detail.
For full details on methods available for downloading files, please see the ASU Research Data Repository Depositor Guide page on downloading files.
For a quick reference on using the download methods mentioned in the ASU Research Data Repository guide, please download and view the ASIST Dataset File Downloads Instructions PDF file.
(2023-01-05)