Cultural Bias in Explainable AI Research: A Systematic Analysis

Authors: Uwe Peters, Mary Carman

JAIR 2024 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Additionally, we systematically reviewed over 200 XAI user studies and found that most studies did not consider relevant cultural variations, sampled only Western populations, but drew conclusions about human-XAI interactions more generally. We also analyzed over 30 literature reviews of XAI studies.
Researcher Affiliation Academia Uwe Peters EMAIL Department of Philosophy, Utrecht University 3512 BL Utrecht, The Netherlands Mary Carman EMAIL Department of Philosophy, University of the Witwatersrand 2050 Johannesburg, South Africa
Pseudocode No The paper describes its methodology for systematic review in prose, outlining steps for paper identification, selection criteria, data extraction, and reliability, but does not present any pseudocode or algorithm blocks.
Open Source Code No The paper states, 'All our data are publicly accessible on an OSF platform here.', referring to the data collected for the review, but does not mention releasing source code for any analysis or methodology described.
Open Datasets Yes All our data are publicly accessible on an OSF platform here.
Dataset Splits No The paper conducts a systematic review of existing literature, which does not involve traditional training/test/validation dataset splits. It analyzes a corpus of papers without partitioning it for machine learning model evaluation.
Hardware Specification No The paper does not provide any specific details about the hardware used to conduct its systematic review and data analysis.
Software Dependencies No The paper does not specify any software dependencies with version numbers used for its systematic review or data analysis.
Experiment Setup No The paper details the methodology for its systematic review, including identification, selection criteria, data extraction, and reliability. However, it does not describe an experimental setup with hyperparameters or system-level training settings, as it is a literature review, not an empirical study training a model.