ISoLA 2024

International Symposium On Leveraging Applications of Formal Methods, Verification and Validation • Crete, Greece

Talk

Team Design Patterns for Meaningful Human Control in Responsible Military Artificial Intelligence

Time: Friday, 1.11

Room: Room 1

Authors: Jurriaan van Diggelen, Hussein Abbass, Christine Boshuizen-van Burken

Abstract: A key requirement for responsible AI in the military is to en- sure meaningful human control (MHC) in operational settings. However, MHC is not a binary concept; it could vary from one operational setting to another depending on context and the competencies of available hu- mans and autonomous systems. We propose and discuss five patterns of MHC and highlight some of their key characteristics. We argue that con- trol is only meaningful when humans have su!cient cognitive resources to perform control functions and when the behavior of autonomous sys- tems in their operational context is su!ciently predictable. By anchoring the MHC patterns in human’s cognition and situation uncertainty, we propose initial requirements for interaction design, testing and validation, and identify gaps in existing scientific understanding.

Slides: Team_Design_Patterns_for_Meaningful_Human_Control_in_Responsible_Military_Artificial_Intelligence-slides.pdf

Paper: Team_Design_Patterns_for_Meaningful_Human_Control_in_Responsible_Military_Artificial_Intelligence-paper.pdf