Aligning Institutional Power and Operational Ability
What it is
This paper proposes a formal framework for analyzing multi-agent AI systems to ensure normative designs align responsibilities with operational capabilities. It uses process-oriented modeling and institutional power concepts to identify and repair incongruences like privilege escalation or unsatisfiable expectations.
The problem it solves
Agentic AI systems often suffer from normative pitfalls where responsibilities are assigned to agents without verifying their operational ability to fulfill them, leading to unsatisfiable expectations or unintended privilege escalation.
What you could use it for
- Designing governance structures for multi-agent AI systems
- Auditing agent architectures for normative pitfalls
- Formalizing the relationship between agent roles and their operational permissions
Why it matters
Provides a rigorous method for preventing social unsustainability in agentic AI by ensuring that agents are only assigned duties they are actually able to perform, thereby reducing risks of privilege escalation and normative failure.
Read the source ↗Back to the library →
Summary written by an AI model from the source; check the source before you rely on it.