AI Ethics: A Bibliometric Analysis, Critical Issues, and Key Gaps

AI ethics has rapidly become a crucial area of academic inquiry. This study provides a thorough bibliometric analysis of AI ethics literature over the past two decades, identifying a three-phase progr

Overview

Authors: Kevin Gao, Andrew Haverly, Sudip Mittal, Jiming Wu, Jingdao Chen

Publication Date: 12 March 2024

Link: https://arxiv.org/pdf/2403.14681

Keywords: AI ethics, bibliomentric analysis, human-centric AI, Collingridge Dilemma, AI transparency, privacy protections, algocracy, superintelligence

Type: Peer-Reviewed Journals/White Papers

Summary

AI ethics has rapidly become a crucial area of academic inquiry. This study provides a thorough bibliometric analysis of AI ethics literature over the past two decades, identifying a three-phase progression: an initial incubation period, a phase centered on embedding human-like attributes into AI, and a final phase prioritizing human-centric AI development. The study highlights seven major ethical challenges, including the Collingridge dilemma, debates over AI's moral status, issues of transparency and explainability, privacy concerns, considerations of justice and fairness, the risks of algocracy and human dependence, and concerns surrounding superintelligence. Additionally, it identifies two key research gaps in AI ethics—the need for large ethics models (LEM) and AI identification mechanisms—while calling for further scholarly exploration in these areas.