From What Ifs to Insights: Counterfactuals in Causal Inference vs. Explainable AI

TL;DR

This paper compares the use and interpretation of counterfactuals in causal inference and explainable AI, highlighting differences and opportunities for cross-disciplinary insights.

Contribution

It introduces a formal definition of counterfactuals applicable to both fields and analyzes their usage, evaluation, and generation to foster cross-fertilization.

Findings

Identifies key differences in counterfactual application between CI and XAI

Proposes a unified formal definition of counterfactuals for both fields

Highlights opportunities for methodological cross-fertilization

Abstract

Counterfactuals play a pivotal role in the two distinct data science fields of causal inference (CI) and explainable artificial intelligence (XAI). While the core idea behind counterfactuals remains the same in both fields--the examination of what would have happened under different circumstances--there are key differences in how they are used and interpreted. We introduce a formal definition that encompasses the multi-faceted concept of the counterfactual in CI and XAI. We then discuss how counterfactuals are used, evaluated, generated, and operationalized in CI vs. XAI, highlighting conceptual and practical differences. By comparing and contrasting the two, we hope to identify opportunities for cross-fertilization across CI and XAI.