9 research outputs found
ASU Hypothesis Pre-registration Spatial Ability
No full textTheory of Mind (ToM) is the ability to infer in others, their attitudes, beliefs, desires, intentions (Butterfill & Apperly, 2013). ToM has gained increased interest recently due to the development of artificial social intelligence (ASI). At a basic level, social intelligence is defined as the ability to work cooperatively with others to achieve a goal (Ford & Tisak, 1983). As such, for ASI to be able to interact with humans effectively, it needs to be able to develop a ToM of the humans in order to predict action or act in an advisory role. To develop ASI that can represent ToM of humans, it is necessary to understand what aspects of the human are needed for the ASI to accurately represent the humans intentions, beliefs, and actions. This study proposes that ASI needs to consider individual spatial ability when developing ToM for human teammates and that ASI needs to consider spatial ability in conjunction with participant experience and overall task complexity in order to improve the performance of ASI agents
Artificial Social Intelligence for Successful Teams (ASIST) Study 3
No full textThe ASIST Study-3 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 withold) 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 -- draw measurements of team state and process from agents called Analytic Components. These take their input from survey responses and behaviors of a three-person team executing an urban search and rescue 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 and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the Defense Advanced Research Projects Agency.
We have partitioned the full dataset into folders that support research in specific areas. A readme file in each folder (e.g., readme_audio.txt) describes the folder's contents in detail.
(1) Data in the studywide folder will be of interest to researchers who conduct any analysis with any data from ASIST Study-2, because these files contain data that describe the study overall, the data used to evaluate AI, or the coding of data.
(2) Data in the surveys folder will be of interest to researchers who study individual differences and their effects on behavior.
(3) Data in the testbedmessages folder will be of interest to researchers who study individual and team behavior or who use any other components of this dataset, because these are machine- and human-readable text (json) records of the state and behaviors of study participants, and of the state of the task environment.
(4) Data in the transcriptions folder will be of interest to researchers who study language use. The audio source of these imperfect machine transcriptions can be found in study video files and audio files.
(5) Data in the audio folder will be of interest to researchers who study language use, or who wish to validate, contextualize, or specify transcriptions, testbed messages, and certain survey data.
(6) Data in the video folder will be of interest to researchers who study machine vision, or who wish to validate, contextualize, or specify transcriptions, testbed messages, and certain survey data.
(7) Data in the analysis folder will be of interest to those seeking examples of analyses developed by ASIST program performers, either to understand the data better, to identify opportunities for further analysis, or to build on analysis code.
(8) Data in the methods folder (and in the studywide folder) will be useful to those seeking to reproduce the human subjects experiment.
</p
Artificial Social Intelligence for Successful Teams (ASIST) Study 2
No full textThe ASIST Study-2 dataset was developed in a human subjects research study designed to assess the capability of artificial intelligence to infer the state and predict the actions of members of a three-person team executing an urban search and rescue task in Minecraft. The data were developed under Contract No. HR001119C0130 to the Defense Advanced Research Projects Agency (DARPA). The dataset comprises approximately 2,100 files and 300GB of data. We have partitioned the full dataset into folders that support research in specific areas. Thus, researchers can more easily download only the files of value to them.
A readme file in each folder (e.g., readme_audio.txt) describes the folder's contents in detail.
(1) Data in the studywide folder will be of interest to researchers who conduct any analysis with any data from ASIST Study-2, because these files contain data that describe the study overall, the data used to evaluate AI, or the coding of data.
(2) Data in the surveys folder will be of interest to researchers who study individual differences and their effects on behavior.
(3) Data in the testbedmessages folder will be of interest to researchers who study individual and team behavior or who use any other components of this dataset, because these are machine- and human-readable text (json) records of the state and behaviors of study participants, and of the state of the task environment.
(4) Data in the transcriptions folder will be of interest to researchers who study language use. The audio source of these imperfect machine transcriptions can be found in study video files and audio files.
(5) Data in the audio folder will be of interest to researchers who study language use, or who wish to validate, contextualize, or specify transcriptions, testbed messages, and certain survey data.
(6) Data in the video folder will be of interest to researchers who study machine vision, or who wish to validate, contextualize, or specify transcriptions, testbed messages, and certain survey data.
</p
Exercises for Artificial Social Intelligence in Minecraft Search and Rescue for Teams
No full textParticipants will engage in a ~10-min screening session to check their capability
to play Minecraft via the Parsec platform. Once determined that they are eligible,
they will be given an intake survey of about 20 minutes to finish on their own.
Submission of the survey will enable them to sign up for the Session 2 Minecraft
experiment. Then, teams of three qualified participants will be scheduled to
participate in a 2-hour experiment that involves searching for victims and
rescuing them in a Minecraft task environment. During that experiment,
participants will receive training videos that introduces the rules of the game and
provides some hands-on experience with the environment before two 17-min
missions. A survey of training knowledge will occur after the training video, and
then a survey to gather reflections about the missions will occur after each
mission.
In the search and rescue task environment, the three participants on each team
will be assigned by the experimenter to one of three roles: Medic, Transporter,
and Engineer. The participants cannot change roles during the missions.
○ The Minecraft environment will simulate a collapsed building. Participants will be
asked to search the building and rescue victims. They will be given a map that
depicts the building’s floor layout prior to the collapse. Participants will need to
remove rubble to clear paths in their search, stabilize victims, and transport
victims from rubble to specific locations. Completion of these tasks produces
rewards that vary in size. Victims stabilized and delivered to the correct zones
count toward the team points: regular victims are 10 points each and critical
victims are 50 points each. Participants will be able to view their accumulated
team score, which is the sum of point rewards for all stabilized victims that are
delivered to the correct zones by all team members.
○ In addition, any participant may enter a threat room whose entryway will then
collapse into rubble when the participant steps on a hidden collapse plate. The
plate is the same color as the rest of the floor. The Engineer, only, can free the
occupant or herself. A threat room entry can collapse repeatedly (with an interval
of 40s since the last collapse was triggered, upon every subsequent trigger of the
collapse plate).
○ All participants will conduct two missions of 17 minutes at equivalent mission
complexity levels in a fixed order. The complexity of the mission is a function of
the complexity of the SAR environment (the building), number of victims, victims’
location by region and by room, the number and location of blockages and threat
rooms (see Appendix Z: Map Design of Victim and Rubble Layout). Perturbations
may be introduced into missions, such as elimination or falsification of some or all
comms (e.g., marker blocks, and/or maps), and rubble falls.
○ The three roles have a unique set of capabilities and knowledge. Specifically:
i. The Medic (Red suit)
1. Walks at a normal speed of 4.317 m/s
2. Can stabilize victims using a medic kit
3. Can diagnose victim injury type when stabilizing victims using an
injury diagnose device (unique knowledge)
4. Can transport victims at the normal speed using a stretcher
5. Has unique knowledge of one's own location (but not the location
of others) through a dynamic map
6. Receives information in the Knowledge Integration Task (KIT,
described below) about the location of some (not all) of the
planned meetings for the day at the time of the collapse through a
text field below the dynamic map.
7. Has unique knowledge of any chat messages from ASI that are
delivered solely to them through chat
ii. The Transporter (Green suit)
1. Can move at the fastest speed of 5.18 m/s (20% faster than
normal
2. Can detect whether there is a victim (at least one regular or a
critical before being moved, awakened, or stabilized) in a room at
each door without entering the room (unique knowledge); this is
done through a victim-detection plate at the door.
3. Can transport victims of all types using a stretcher at the fastest
speed
4. Can help the medic to wake up a critical victim
5. Has unique knowledge of one's own location (but not the location
of others) through a dynamic map.
6. Receives KIT information about the number of attendees for some
(not all) of the planned meetings for the day at the time of the
collapse through a text field below the dynamic map.
7. Has unique knowledge of any chat messages from ASI that are
delivered solely to them through chat
iii. The Engineer (Blue suit)
1. Can move at a slow speed of 3.669 m/s (15% slower than normal)
2. Can remove rubble using a hammer
3. Can transport victims at a slow speed
4. Can see the threat rooms on the dynamic map (unique
knowledge)
5. Has unique knowledge of one's own location (but not the location
of others) through a dynamic map
6. Receives KIT information about the severity of some (but not all)
rooms where injuries have happened through the text field below
the dynamic map.
7. Has unique knowledge of any chat messages from ASI that are
delivered solely to them through chat.
○ The tools of all roles are impervious to wear4.
○ The three members of a team will each have a dynamic map of the mission
space (i.e., the map is updated automatically in real time). These maps depict
some information in common to all participants, specifically building structure,
and the location and type of marker blocks (but not rubble location or victim
location).
○ Knowledge Integration Task (KIT): Each mission has one KIT. Each member
receives part of the KIT information that, when combined, rewards teamwork,
specifically information sharing. Successfully solving the KIT will enable
participants to identify all three rooms that have critical victims, especially one of
them that has 3 critical victims. Participants will be given 1 minute to read a
mission brief slide containing KIT information before the mission starts. After the
roles are selected and a mission starts, the KIT information will also be displayed
below the dynamic map for the entire mission. That information is as follows:
i. Only the Medic receives information about the location of some (not all) of
the planned meetings for the day at the time of the collapse.
ii. Only the Transporter receives information about the number of attendees
for some (not all) of the planned meetings for the day at the time of the
collapse.
iii. Only the Engineer receives information about the severity of some (not
all) rooms where injuries have happened.
○ The three participants have the same set of marker blocks, which are designed to
communicate information about the game and indicate the identity (Medic-Red,
Transporter-Green, or Engineer-Blue) of the author. Marker blocks laid in the
game will appear on the maps of all players dynamically. These marker blocks
indicate: Regular victim here, Critical victim here, No victim here, Victim injury
type A, Victim injury type B, Victim injury type C, Rubble here, Threat room here,
and help me here5.
○ The program evaluation team (TA3) will apply approximately 30 measures that
provide quantitative or qualitative evidence concerning these claims for the
ASIST program: that social science implemented in Analytic Components drives
ASI MToT, which in turn drives ASI interventions, which influence team process,
and potentially influence mission effects. That evaluation will contrast the eight
between-teams experimental conditions (the no-advisor baseline, human advisor
benchmark, and each of the six ASI advisors). See the Evaluation section of this
document for detail
Exercises for Artificial Social Intelligence in Minecraft Search and Rescue for Teams
No full textParticipants will engage in a ~10-min screening session to check their capability
to play Minecraft via the Parsec platform. Once determined that they are eligible,
they will be given an intake survey of about 20 minutes to finish on their own.
Submission of the survey will enable them to sign up for the Session 2 Minecraft
experiment. Then, teams of three qualified participants will be scheduled to
participate in a 2-hour experiment that involves searching for victims and
rescuing them in a Minecraft task environment. During that experiment,
participants will receive training videos that introduces the rules of the game and
provides some hands-on experience with the environment before two 17-min
missions. A survey of training knowledge will occur after the training video, and
then a survey to gather reflections about the missions will occur after each
mission.
In the search and rescue task environment, the three participants on each team
will be assigned by the experimenter to one of three roles: Medic, Transporter,
and Engineer. The participants cannot change roles during the missions.
○ The Minecraft environment will simulate a collapsed building. Participants will be
asked to search the building and rescue victims. They will be given a map that
depicts the building’s floor layout prior to the collapse. Participants will need to
remove rubble to clear paths in their search, stabilize victims, and transport
victims from rubble to specific locations. Completion of these tasks produces
rewards that vary in size. Victims stabilized and delivered to the correct zones
count toward the team points: regular victims are 10 points each and critical
victims are 50 points each. Participants will be able to view their accumulated
team score, which is the sum of point rewards for all stabilized victims that are
delivered to the correct zones by all team members.
○ In addition, any participant may enter a threat room whose entryway will then
collapse into rubble when the participant steps on a hidden collapse plate. The
plate is the same color as the rest of the floor. The Engineer, only, can free the
occupant or herself. A threat room entry can collapse repeatedly (with an interval
of 40s since the last collapse was triggered, upon every subsequent trigger of the
collapse plate).
○ All participants will conduct two missions of 17 minutes at equivalent mission
complexity levels in a fixed order. The complexity of the mission is a function of
the complexity of the SAR environment (the building), number of victims, victims’
location by region and by room, the number and location of blockages and threat
rooms (see Appendix Z: Map Design of Victim and Rubble Layout). Perturbations
may be introduced into missions, such as elimination or falsification of some or all
comms (e.g., marker blocks, and/or maps), and rubble falls.
○ The three roles have a unique set of capabilities and knowledge. Specifically:
i. The Medic (Red suit)
1. Walks at a normal speed of 4.317 m/s
2. Can stabilize victims using a medic kit
3. Can diagnose victim injury type when stabilizing victims using an
injury diagnose device (unique knowledge)
4. Can transport victims at the normal speed using a stretcher
5. Has unique knowledge of one's own location (but not the location
of others) through a dynamic map
6. Receives information in the Knowledge Integration Task (KIT,
described below) about the location of some (not all) of the
planned meetings for the day at the time of the collapse through a
text field below the dynamic map.
7. Has unique knowledge of any chat messages from ASI that are
delivered solely to them through chat
ii. The Transporter (Green suit)
1. Can move at the fastest speed of 5.18 m/s (20% faster than
normal
2. Can detect whether there is a victim (at least one regular or a
critical before being moved, awakened, or stabilized) in a room at
each door without entering the room (unique knowledge); this is
done through a victim-detection plate at the door.
3. Can transport victims of all types using a stretcher at the fastest
speed
4. Can help the medic to wake up a critical victim
5. Has unique knowledge of one's own location (but not the location
of others) through a dynamic map.
6. Receives KIT information about the number of attendees for some
(not all) of the planned meetings for the day at the time of the
collapse through a text field below the dynamic map.
7. Has unique knowledge of any chat messages from ASI that are
delivered solely to them through chat
iii. The Engineer (Blue suit)
1. Can move at a slow speed of 3.669 m/s (15% slower than normal)
2. Can remove rubble using a hammer
3. Can transport victims at a slow speed
4. Can see the threat rooms on the dynamic map (unique
knowledge)
5. Has unique knowledge of one's own location (but not the location
of others) through a dynamic map
6. Receives KIT information about the severity of some (but not all)
rooms where injuries have happened through the text field below
the dynamic map.
7. Has unique knowledge of any chat messages from ASI that are
delivered solely to them through chat.
○ The tools of all roles are impervious to wear4.
○ The three members of a team will each have a dynamic map of the mission
space (i.e., the map is updated automatically in real time). These maps depict
some information in common to all participants, specifically building structure,
and the location and type of marker blocks (but not rubble location or victim
location).
○ Knowledge Integration Task (KIT): Each mission has one KIT. Each member
receives part of the KIT information that, when combined, rewards teamwork,
specifically information sharing. Successfully solving the KIT will enable
participants to identify all three rooms that have critical victims, especially one of
them that has 3 critical victims. Participants will be given 1 minute to read a
mission brief slide containing KIT information before the mission starts. After the
roles are selected and a mission starts, the KIT information will also be displayed
below the dynamic map for the entire mission. That information is as follows:
i. Only the Medic receives information about the location of some (not all) of
the planned meetings for the day at the time of the collapse.
ii. Only the Transporter receives information about the number of attendees
for some (not all) of the planned meetings for the day at the time of the
collapse.
iii. Only the Engineer receives information about the severity of some (not
all) rooms where injuries have happened.
○ The three participants have the same set of marker blocks, which are designed to
communicate information about the game and indicate the identity (Medic-Red,
Transporter-Green, or Engineer-Blue) of the author. Marker blocks laid in the
game will appear on the maps of all players dynamically. These marker blocks
indicate: Regular victim here, Critical victim here, No victim here, Victim injury
type A, Victim injury type B, Victim injury type C, Rubble here, Threat room here,
and help me here5.
○ The program evaluation team (TA3) will apply approximately 30 measures that
provide quantitative or qualitative evidence concerning these claims for the
ASIST program: that social science implemented in Analytic Components drives
ASI MToT, which in turn drives ASI interventions, which influence team process,
and potentially influence mission effects. That evaluation will contrast the eight
between-teams experimental conditions (the no-advisor baseline, human advisor
benchmark, and each of the six ASI advisors). See the Evaluation section of this
document for detail
Exercises for Artificial Social Intelligence in Minecraft Search and Rescue for Teams
No full textParticipants will engage in a ~10-min screening session to check their capability
to play Minecraft via the Parsec platform. Once determined that they are eligible,
they will be given an intake survey of about 20 minutes to finish on their own.
Submission of the survey will enable them to sign up for the Session 2 Minecraft
experiment. Then, teams of three qualified participants will be scheduled to
participate in a 2-hour experiment that involves searching for victims and
rescuing them in a Minecraft task environment. During that experiment,
participants will receive training videos that introduces the rules of the game and
provides some hands-on experience with the environment before two 17-min
missions. A survey of training knowledge will occur after the training video, and
then a survey to gather reflections about the missions will occur after each
mission.
In the search and rescue task environment, the three participants on each team
will be assigned by the experimenter to one of three roles: Medic, Transporter,
and Engineer. The participants cannot change roles during the missions.
○ The Minecraft environment will simulate a collapsed building. Participants will be
asked to search the building and rescue victims. They will be given a map that
depicts the building’s floor layout prior to the collapse. Participants will need to
remove rubble to clear paths in their search, stabilize victims, and transport
victims from rubble to specific locations. Completion of these tasks produces
rewards that vary in size. Victims stabilized and delivered to the correct zones
count toward the team points: regular victims are 10 points each and critical
victims are 50 points each. Participants will be able to view their accumulated
team score, which is the sum of point rewards for all stabilized victims that are
delivered to the correct zones by all team members.
○ In addition, any participant may enter a threat room whose entryway will then
collapse into rubble when the participant steps on a hidden collapse plate. The
plate is the same color as the rest of the floor. The Engineer, only, can free the
occupant or herself. A threat room entry can collapse repeatedly (with an interval
of 40s since the last collapse was triggered, upon every subsequent trigger of the
collapse plate).
○ All participants will conduct two missions of 17 minutes at equivalent mission
complexity levels in a fixed order. The complexity of the mission is a function of
the complexity of the SAR environment (the building), number of victims, victims’
location by region and by room, the number and location of blockages and threat
rooms (see Appendix Z: Map Design of Victim and Rubble Layout). Perturbations
may be introduced into missions, such as elimination or falsification of some or all
comms (e.g., marker blocks, and/or maps), and rubble falls.
○ The three roles have a unique set of capabilities and knowledge. Specifically:
i. The Medic (Red suit)
1. Walks at a normal speed of 4.317 m/s
2. Can stabilize victims using a medic kit
3. Can diagnose victim injury type when stabilizing victims using an
injury diagnose device (unique knowledge)
4. Can transport victims at the normal speed using a stretcher
5. Has unique knowledge of one's own location (but not the location
of others) through a dynamic map
6. Receives information in the Knowledge Integration Task (KIT,
described below) about the location of some (not all) of the
planned meetings for the day at the time of the collapse through a
text field below the dynamic map.
7. Has unique knowledge of any chat messages from ASI that are
delivered solely to them through chat
ii. The Transporter (Green suit)
1. Can move at the fastest speed of 5.18 m/s (20% faster than
normal
2. Can detect whether there is a victim (at least one regular or a
critical before being moved, awakened, or stabilized) in a room at
each door without entering the room (unique knowledge); this is
done through a victim-detection plate at the door.
3. Can transport victims of all types using a stretcher at the fastest
speed
4. Can help the medic to wake up a critical victim
5. Has unique knowledge of one's own location (but not the location
of others) through a dynamic map.
6. Receives KIT information about the number of attendees for some
(not all) of the planned meetings for the day at the time of the
collapse through a text field below the dynamic map.
7. Has unique knowledge of any chat messages from ASI that are
delivered solely to them through chat
iii. The Engineer (Blue suit)
1. Can move at a slow speed of 3.669 m/s (15% slower than normal)
2. Can remove rubble using a hammer
3. Can transport victims at a slow speed
4. Can see the threat rooms on the dynamic map (unique
knowledge)
5. Has unique knowledge of one's own location (but not the location
of others) through a dynamic map
6. Receives KIT information about the severity of some (but not all)
rooms where injuries have happened through the text field below
the dynamic map.
7. Has unique knowledge of any chat messages from ASI that are
delivered solely to them through chat.
○ The tools of all roles are impervious to wear4.
○ The three members of a team will each have a dynamic map of the mission
space (i.e., the map is updated automatically in real time). These maps depict
some information in common to all participants, specifically building structure,
and the location and type of marker blocks (but not rubble location or victim
location).
○ Knowledge Integration Task (KIT): Each mission has one KIT. Each member
receives part of the KIT information that, when combined, rewards teamwork,
specifically information sharing. Successfully solving the KIT will enable
participants to identify all three rooms that have critical victims, especially one of
them that has 3 critical victims. Participants will be given 1 minute to read a
mission brief slide containing KIT information before the mission starts. After the
roles are selected and a mission starts, the KIT information will also be displayed
below the dynamic map for the entire mission. That information is as follows:
i. Only the Medic receives information about the location of some (not all) of
the planned meetings for the day at the time of the collapse.
ii. Only the Transporter receives information about the number of attendees
for some (not all) of the planned meetings for the day at the time of the
collapse.
iii. Only the Engineer receives information about the severity of some (not
all) rooms where injuries have happened.
○ The three participants have the same set of marker blocks, which are designed to
communicate information about the game and indicate the identity (Medic-Red,
Transporter-Green, or Engineer-Blue) of the author. Marker blocks laid in the
game will appear on the maps of all players dynamically. These marker blocks
indicate: Regular victim here, Critical victim here, No victim here, Victim injury
type A, Victim injury type B, Victim injury type C, Rubble here, Threat room here,
and help me here5.
○ The program evaluation team (TA3) will apply approximately 30 measures that
provide quantitative or qualitative evidence concerning these claims for the
ASIST program: that social science implemented in Analytic Components drives
ASI MToT, which in turn drives ASI interventions, which influence team process,
and potentially influence mission effects. That evaluation will contrast the eight
between-teams experimental conditions (the no-advisor baseline, human advisor
benchmark, and each of the six ASI advisors). See the Evaluation section of this
document for detail
Artificial Social Intelligence for Successful Teams (ASIST) Study 4 Dragon Testbed Dataset
No full textArtificial 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.</p
ASIST Study 2 June 2021 Exercises for Artificial Social Intelligence in Minecraft Search and Rescue for Teams
No full textParticipants will participate in a 1-hour session to install required software appropriately and fill out independent surveys. Then, teams of three qualified participants will participate in a 2.5-hour session to complete an experiment in which they search for victims and rescue them in a Minecraft task environment. Participants will receive training that introduces the rules of the game and provides some hands-on experience with the environment. In the search and rescue task environment, the three participants on each team may choose one of three roles -- Medical Specialist (medic), Search Specialist (searcher), or Heavy Equipment Specialist (engineer) -- and change those roles during the mission
ASIST TA3 Study 2 Results
No full textParticipants will participate in a 1-hour session to install required
software appropriately and fill out independent surveys. Then, teams of three
qualified participants will participate in a 2.5-hour session to complete an
experiment in which they search for victims and rescue them in a Minecraft task
environment. Participants will receive training that introduces the rules of the
game and provides some hands-on experience with the environment. In the
search and rescue task environment, the three participants on each team may
choose one of three roles -- Medical Specialist (medic), Search Specialist
(searcher), or Heavy Equipment Specialist (engineer) -- and change those roles
during the mission
