A Human-Centred Process Model for Explainable AI

EasyChair Preprint 7332

Abstract

With the increase in inscrutable Artificial Intelligence systems being used to support human decision makers, there has been much interest in what it means for these systems to provide ‘explanation’. In this paper, the concern is with applying a simple formalism that can express a minimal set of features that can be used to define an explanation. It is argued that few contemporary AI systems support this minimal set. Advice is provided on how future developments in explainable AI systems could adhere to this minimal set.

Keyphrases: Explainable AI, XAI, explanation

@booklet{EasyChair:7332,
  author    = {Chris Baber and Emily McCormick and Ian Apperly},
  title     = {A Human-Centred Process Model for Explainable AI},
  howpublished = {EasyChair Preprint 7332},
  year      = {EasyChair, 2022}}