EVALUATING AI DETECTION TOOLS FOR ACADEMIC INTEGRITY IN HIGHER EDUCATION

Abstract

The advent of artificial intelligence (AI) in academic settings has presented educators with unprecedented challenges in maintaining academic integrity.This study examines the efficacy of three AI-powered text-checking services-Quillbot, HiveModeration, and ZeroGPT-in detecting AI-generated content and alterations in academic texts.The investigation employs five text variants: text generated by ChatGPT, original pre-AI era text, AI-rephrased text (Quillbot), Quillbot-rephrased original text, and edited (minor edits) original text by DeepL and Grammarly services.All three services exhibit 100% accuracy in detecting AI-generated text while failing to identify AI signs in original English text.Surprisingly, when examining the original Ukrainian text, ZeroGPT detects AI signs with 100% accuracy.Quillbot's paraphrasing substantially alters the text, leading to varied detection rates across the services.Notably, AI-induced alterations influence detection results, exemplified by Quillbot showing high detection rates in ChatGPTgenerated and original text (79% and 60% respectively).Lastly, editing original text by DeepL and Grammarly does not trigger AI detection.In conclusion, this study offers several key findings and recommendations.Firstly, while AI detection tools excel at identifying machine-generated text, they exhibit vulnerability to bypassing through paraphrasing and specific textual styles.Secondly, false positive results underscore the unreliability of these services, highlighting the need for cautious interpretation.Thirdly, education on ethical AI use is imperative, distinguishing 5( 33) 2024 ( , , , , ) 971 between permissible and impermissible behaviors and integrating AI tools into assignments transparently.Fourthly, adapting educational systems to accommodate AI tools' use and fostering collaborative, interdisciplinary projects can enhance students' AI literacy and ethical awareness.Finally, comprehensive policies outlining ethical AI usage guidelines are essential to navigate the ethical complexities of AI integration in education effectively.

Ähnliche Arbeiten