Home // DIGITAL 2022, Advances on Societal Digital Transformation // View article

War-Gaming Needs Argument-Justified AI More Than Explainable AI

Authors:
John Licato

Keywords: explainable AI, argumentation, AI, wargaming

Abstract:
I argue that a planning agent in a societal- or war-gaming environment, whether that agent is a sole AI or part of a human-AI team, should behave in a way that is more than just explainable. Rather, its actions should be argument-justified; i.e., it must produce as justification of its actions the equivalent of an argument graph demonstrating how its choice is superior to, and fairly considers, the strongest possible arguments for a sufficient number of alternative choices. Although argument-justified AI might be considered a subset of interpretable AI, the requirement that a qualified argument graph be part of the model's output imparts multiple desirable properties over alternatives, namely: trustworthiness, understandability, persuasiveness, thoroughness, and others.

Pages: 1 to 8

Copyright: Copyright (c) IARIA, 2022

Publication date: November 13, 2022

Published in: conference

ISBN: 978-1-68558-014-8

Location: Valencia, Spain

Dates: from November 13, 2022 to November 17, 2022