•  
  •  
 

Research Highlights

  • The Problem: AI plagiarism detection systems create an "integrity paradox" by granting ethical authority to algorithms that conflate statistical similarity with misconduct and offer automated rewrites that replace human authorship. 

  • The Method: Authors Elliott Ostler and Jeanne Surface utilized a case study of a 1,000-word original ethics editorial submitted to an online AI plagiarism checker to analyze "epistemic substitution" and the collapse of evaluator/author roles. 

  • Quantitative Finding: The AI software labeled a 1,000-word original manuscript as 40.7% "non-original"; academic journals often set an ethical threshold for similarity at 15%. 

  • Qualitative Finding: Plagiarism tools measure textual similarity rather than intent or expertise; disciplinary fluency in specialized fields naturally produces higher similarity scores; algorithmic "fixing" buttons convert human expression into AI-generated text, shifting accountability without true consent. 

  • Finding: Ethical oversight must be reclaimed by human reviewers through contextual, domain-specific analysis and explicit training on the limitations of similarity reports.

Abstract

This article critiques the ethical authority granted to AI plagiarism detection systems, using humor to expose a serious contradiction in these algorithms. Through a case in which an original ethics manuscript is labeled 40.7% “non-original,” the authors argue that similarity scores are mistaken for judgments of misconduct. They introduce the concept of “epistemic substitution” to describe how tools that both judge originality and offer AI rewrites replace human authorship while mandating human accountability. The paper contends that this collapse of roles undermines scholarly integrity, conflates disciplinary fluency with plagiarism, and shifts ethical responsibility from humans to procedures. The authors call for contextual, human-centered oversight of AI-assisted evaluation.

Download

Share

COinS