AI Norm-Guided Planning with Dynamic Norms
A new paper from computer science researchers presents an approach for AI agents to plan actions guided by dynamically changing human norms. The work introduces a defeasible calculus to resolve normative conflicts and uses norms as guard rails on plans. The approach is demonstrated theoretically with formal proofs and empirically with an AI agent called SocialBot on a natural language dialogue task. The paper addresses the gap in norm-guided planning, which previously only considered static norms within artificial agent communities, ignoring the dynamic nature of norms in human-AI settings.
Key facts
- The paper is titled 'Reasoning and Planning with Dynamically Changing Norms'.
- It is categorized under Computer Science > Artificial Intelligence.
- The approach uses a defeasible calculus for resolving normative conflicts.
- Norms are used as guard rails on plans.
- Theoretical demonstration includes formal proofs.
- Empirical demonstration uses an AI agent called SocialBot.
- The task is natural language dialogue.
- The work addresses dynamic norms in human-AI settings.
Entities
Institutions
- arXiv