Concurrent Session #9: Successfully Identifying AI Submissions for Short-Essay Exam Questions

Alternative Title

Successfully Identifying Artificial Intelligence (AI) Submissions for Short-Essay Exam Questions

Contributor

University of Central Florida. Faculty Center for Teaching and Learning; University of Central Florida. Division of Digital Learning; Teaching and Learning with AI Conference (2023 : Orlando, Fla.)

Location

Cape Florida A

Start Date

25-9-2023 2:45 PM

End Date

25-9-2023 3:00 PM

Publisher

University of Central Florida Libraries

Keywords:

AI detection; Short essay evaluation; ChatGPT responses; Exam integrity; Educational technology

Subjects

Artificial intelligence--Educational applications; Artificial intelligence--Experiments; Grading and marking (Students)--Computer-assisted instruction; Artificial intelligence--Study and teaching; Academic writing--Examinations

Description

Artificial Intelligence (AI) tools such as ChatGPT can easily produce short essay responses to the prompts used in exam questions. To determine whether the AI responses could be successfully identified, 25 actual student responses to an exam question were combined with an equal number of responses generated by ChatGPT and evaluated by both a human instructor and the GPTZero tool. The GPTZero tool computes metrics on the word use in text which it uses to determine whether the text was generated entirely by a human, entirely by an AI, or by a human with AI help.

Language

eng

Type

Presentation

Rights Statement

All Rights Reserved

Audience

Faculty, Educators, Administrators

This document is currently not available here.

Share

COinS
 
Sep 25th, 2:45 PM Sep 25th, 3:00 PM

Concurrent Session #9: Successfully Identifying AI Submissions for Short-Essay Exam Questions

Cape Florida A

Artificial Intelligence (AI) tools such as ChatGPT can easily produce short essay responses to the prompts used in exam questions. To determine whether the AI responses could be successfully identified, 25 actual student responses to an exam question were combined with an equal number of responses generated by ChatGPT and evaluated by both a human instructor and the GPTZero tool. The GPTZero tool computes metrics on the word use in text which it uses to determine whether the text was generated entirely by a human, entirely by an AI, or by a human with AI help.