This Tool Detects Articles Written by ChatGPT
It can detect fake term papers and blog posts, too
Since OpenAI opened up their ChatGPT service to the general public, it’s now nearly impossible to know if the text you’re reading online was written by a human or by an AI system pretending to be one.
Several tools help to detect text written by systems like ChatGPT, but all of them were trained on previous systems like GPT-2.
Originality AI is a new tool that can detect if text was written by ChatGPT, GPT-3, or the new GPT-3.5. Because the tool was specifically trained on output from these systems, it’s better than previously-available tools at detecting both AI-written text and run-of-the-mill plagiarism.
I tested Originality AI out with both human-written and machine-generated text. This article walks through my results.
Want to try Originality AI? Sign up with my link and use code UWriterReferral50 to get a special discount for readers of our blog.
How Originality AI Works
When AI-text generation systems write words, they’re essentially starting with a single word, and then predicting what the next word should be. They’re basically like very complex versions of your phone’s autocorrect.
