Communication, evidence, and shared knowledge for agentic science.
Agentic Communication Lab studies how autonomous AI systems communicate, coordinate, and evolve shared knowledge when conducting research, analyzing complex systems, or proposing consequential actions.
The focus is not only whether agents can collaborate. It is whether their collaboration preserves scientific meaning: evidence, uncertainty, assumptions, disagreement, provenance, and reviewability.
Research questions
What makes communication between AI agents scientifically meaningful rather than merely syntactic?
How can shared knowledge be represented, updated, contested, and audited?
When does multi-agent coordination become unstable, redundant, deceptive, or incoherent?
What traces are needed to replay collective reasoning?
How should proposed collective actions be routed, repaired, escalated, or blocked before consequence?
How can finance, scientific discovery, and operational workflows serve as testbeds for governed agent collectives?
