Multi-Agent Architecture
open-problem
ClaimThe full system will be a collection of AI agents, adaptive controllers, and data-collecting tools. How multiple specialized agents coordinate, share state estimates, and maintain caring orientation across a distributed system is unspecified.
The problem
The machine is not a single AI. It is more likely a multi-agent AI system with many simple adaptive controllers, sensors, and actuators, and a hierarchy of agent orchestration eventually leading to a single output for the user.
Open questions
- How do multiple specialized agents coordinate?
- How do they share state estimates?
- How is the caring orientation maintained across a distributed system?
- How is the human-as-controller topology preserved when the "machine" is actually many machines?
- How are the caring invariants (I1-I6) enforced across agent boundaries?
Status
Architecture at a level beyond the initial specification. Essential for long-term development but not required for the experimental program (E1-E4), which can operate with a simpler single-agent or small-ensemble system.