To delve or not to delve: is that the question? AI detection tools

John Glenis

Research output: Unpublished conference presentation (presented paper, abstract, poster)Conference presentation (not published in journal/proceeding/book)Academic research

Abstract

This interactive demonstration aims to investigate to what extent tools like Originality AI, QuillBot, GPTZero, CopyLeaks, Sapling and Turnitin can effectively detect AI-generated content. It also looks at how manipulating a text - making grammatical and lexical adjustments - can affect the performance of an AI detection tool. Human intervention as well as the employment of AI humanizer tools are both examined for their efficiency in bypassing detection.
Moreover, this demonstration explores to what extent teachers alone can detect AI-generated text and other forms of academic dishonesty. It promotes a blended approach as the most reliable method, and suggests practical tips towards such a multi-dimensional approach, as well as possible adjustments in the ways students are assessed and rubrics developed. Finally, it argues that academic institutions which accept, integrate and facilitate the ethical use of AI tools should also assess it as an academic skill.
Original languageEnglish
Publication statusNot published / presented only - 22 May 2025
EventAHKLC Symposium 2025: Repositioning Language Education in the Age of AI - Hong Kong Polytechnic University, Kowloon, Hong Kong
Duration: 22 May 202523 May 2025
https://events.polyu.edu.hk/ahklc2025/home

Forum/Symposium

Forum/SymposiumAHKLC Symposium 2025
Country/TerritoryHong Kong
CityKowloon
Period22/05/2523/05/25
Internet address

Fingerprint

Dive into the research topics of 'To delve or not to delve: is that the question? AI detection tools'. Together they form a unique fingerprint.

Cite this