The need to Develop Cooperative AI Agents and Robots
When human professionals
undertake a program or project and execute it, several experts, specialists, and
generalists do various tasks as a team. Each of them knows what they can
do what they need to delegate to other professionals. They know who those other
persons are or know how to find them and do so. Work is assigned along with
terms of reference and expectations. People communicate during the performance
of the tasks to get clarifications, to issue changes, to report status, and so
on. They evaluate progress and re-plan as needed.
It is too much to expect one AI
system (or AI agent) to have the capability (currently) to do all these
distinct tasks in multitudinous disciplines. Therefore, what we need to develop
are hierarchical AI systems with attributes similar to those that humans
possess as described above. We may call them Cooperative AI Agents, which know
how to work as a team and accomplish a task.
Here is a presentation describing the concept briefly.
No comments:
Post a Comment