A Taxonomy of Questions for Critical Reflection in Machine-Assisted Decision-Making

TL;DR

This paper addresses the problem of automation-induced under-engagement in machine-assisted decision-making by introducing a taxonomy of 10 question types that promote reflection and cognitive engagement. Built from a synthesis of Socratic questioning, Bloom's taxonomy, and a real-world XAI question bank, the method identifies what to reflect on and how to formulate questions, starting with clinical decision-making and tested for transferability to education. The main contributions include a structured approach to generate questions that stimulate critical thinking, a framework for presenting these questions (including a proposed Reflection Machine), and empirical feedback from medical and educational experts that supports relevance and potential impact. The study aims to mitigate overreliance and support responsible human oversight in high-stakes domains, aligning with the European AI Act’s emphasis on maintaining human oversight and awareness of automation bias.

Abstract

Decision-makers run the risk of relying too much on machine recommendations, which is associated with lower cognitive engagement. Reflection has been shown to increase cognitive engagement and improve critical thinking and therefore decision-making. Questions are a means to stimulate reflection, but there is a research gap regarding the systematic creation and use of relevant questions for machine-assisted decision-making. We therefore present a taxonomy of questions aimed at promoting reflection and cognitive engagement in order to stimulate a deliberate decision-making process. Our taxonomy builds on the Socratic questioning method and a question bank for explainable AI. As a starting point, we focus on clinical decision-making. Brief discussions with two medical and three educational researchers provide feedback on the relevance and expected benefits of our taxonomy. Our work contributes to research on mitigating overreliance in human-AI interactions and aims to support effective human oversight as required by the European AI Act.

Figures (1)

• Figure 1: Our elements for reflection in machine-assisted decision-making in the middle column (see section \ref{['taxonomy']}) are based on the taxonomy for Socratic questions Paul2019, in the left column, and the XAI question bank Liao2022, in the right column.