Skip to Main Content

Artificial Intelligence: Detecting AI Generated Text

Resources for Artificial Intelligence in Higher Education, ChatGPT, Dall-E, and others.

 

Please keep in mind that there is not a tool currently on the market that is 100% reliable. The tools mentioned on this page have been praised among different tech related news websites, blogs, and videos. The OCC Library doesn't endorse any software, hardware, or other technologies mentioned on this guide.

Tools for Detecting AI

Tips for Detecting AI Generated Content

While there isn't a perfect tool or resource to accurately determine whether content was written by AI without errors, there are a few red flags to look for:

  1. Repetitive answers
    • ChatGPT will generate very similar answers with very similar language. If you're starting to see the same construction, examples, or reasoning in a number of papers, this could be a sign that multiple students are using AI generated content
  2. Past information
    • ChatGPT does not have information on current events. It only pulls information from 2021 or earlier.
  3. A lot of fluffy language
    • ChatGPT uses a lot of long-winded answers from multiple directions. OpenAI stated that "the model is often excessively verbose and overuses certain phrases".
  4. Plausible but not factual
    • OpenAI admits that it "sometimes writes plausible-sounding but incorrect or nonsensical answers".
  5. Bias towards neutrality
    • ChatGPT is generally neutral and informative and doesn't usually give an opinion or preference.