Human-Agent Teamwork
Joint Activity
Interpredictability
- Ability to predict what others will do
- Not just actions, but the coordination itself
- The suitcase example
Common ground
- Mutual knowledge, beliefs, and other skills and capabilities
Directablity
- Modify the actions when conditions change
- Respectiveness of each participants
Theory of mind
Humans are inclined to ascribe agency characteristics, like intentions to (self-)moving objects. A theory of mind helps to improve the social world, i.e. it helps to understand, explain, predict or manipulate others' behavior.
Shared mental model
Knowledge shared by team members. It is used to describe, explain and predict team behavior and allows team members to coordinate with each other.
Types of knowledge to share
- Task-related vs Team-related knowledge
- Facts vs Intentions
Different team structures
- Teams with key members (managers, that know more than others)
- Teams with subteams (each responsible for their own task)
- Teams with changing team members (where it is important to avoid loss of knowledge)
We are a team with key members because key members in our team know more about the project. Actually, it is a good team structure because it is better to have someone who knows the whole project and divides tasks for other team members.
How much is the XWiki helpful? Who is responsible for updating the XWiki?
XWiki is really helpful because we could know what others are doing and the process of the project. In our team, we are responsible for updating different parts of XWiki.
Transactive memory system
We should face a transactive memory system (TMS) that is distributed across different team members. Each of the team members has knowledge about who knows what.
Task Management
Coordination
- Manage dependencies between activities
- Temporal sequencing of behaviors among team members
- To act in a harmonious combination
In our team, we maintain an online sheet to track all the tasks that need to be accomplished. We put specific team members' names to indicate who is going to do the task. Besides, we use different colors to indicate the state of tasks. For example, red means "not done", orange means "needs to be improved", blue means "fine but can be improved", and green means "good". In this way, we know what each other is doing and the progress of tasks.