AN ARCHITECTURE FOR NORMATIVE, EXPLAINABLE, AND JUSTIFIED AGENCY
Abstract
Before intelligent systems can gain widespread acceptance in various human-machine teams, they must act in accordance with the rules of the teams and they must communicate their decision making to humans in ways that convince team members that they share the same goals. We want intelligent systems to exhibit an ability to justify its decisions. The aim is to develop a computational theory of intelligent systems that can explain their behavior in understandable terms that relate to human social norms. This work will study various architectures for different classes of AI performance.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Aug 12, 2021
- Source ID
- FA95502010130
Entities
People
- Larry John Leifer
Organizations
- Air Force Office of Scientific Research
- Stanford University
- United States Air Force