Artificial Intelligence and ChatGPT: Academic Integrity & Ethics
Publisher Statements
- ElsevierThe use of AI and AI-assisted writing technologies in scientific writing
Academic Response to ChatGPT
- What Students Are Saying About ChatGPTThe Learning Network, New York Times
- Using AI to write scholarly publicationsMohammed Hosseini, Lisa M. Rasmussen, & David B. Resnik
Citing Generative AI
In Defense of ChatGPT
Pros:
- Increase in student engagement
- smart feedback
- 24/7 help
- cost effective
Objections to ChatGPT
Cons:
- lack of human interaction
- ChatGPT has limited knowledge
- dependence on technology for learning
- plagiarism/cheating
Detecting ChatGPT?
- Open AI is working on a new AI classifier to detect AI-written text
- Limitations:
- not fully reliable
- unreliable for shorter texts
- best used for English texts (not as good with other langages/code)
- AI-written text can be edited to trick it
- Limitations:
- GPT Zero
- created by a Princeton University student
- uses perplexity and burstiness to detect text
- "We pitted ChatGPT against tools for detecting AI-written text, and the results are troubling"
- an article published by The Conversation comparing different ways to detect AI-written text
- "Most sites claiming to catch AI-written text fail spectacularly"
- an article published by the Tech Crunch about the limitations of AI detection
Web Address: library.rice.edu | Physical Address: 6100 Main Street, Houston, Texas 77005 | Mailing Address: MS-44, P.O. Box 1892, Houston, Texas 77251-1892 | Phone: 713-348-5698 | © 2014 Rice University