Edward Tian, a computer science major at Princeton University, has created an app called GPTZero that can determine if a text was authored by a human or by the artificial intelligence tool ChatGPT.
Tian stated in a recent tweet that the app’s algorithm can “quickly and efficiently detect whether an essay is ChatGPT or human written.” GPTZero evaluates a passage’s “perplexity” and “burstiness” to determine whether it was written by a bot. Perplexity refers to how random the text in a sentence is, as well as whether the way a sentence is constructed is unusual or surprising to the app. Burstiness compares these sentences to determine their similarity. Human writing has more burstiness, which means it contains more sentence variation.
His motivation for developing the bot was to combat what he perceives as an increase in AI plagiarism. There have been reports of students using ChatGPT’s breakthrough language model to pass off AI-written assignments as their own since its release in late November.
“There’s so much ChatGPT hype going around. is this and that written by AI? we as humans deserve to know!” Tian wrote in a tweet introducing GPTZero.
Tian released a few of proof-of-concept videos explaining GPTZero’s capabilities on January 2nd. It recognised ChatGPT as the author of a Facebook post after first determining that a human authored a New Yorker article.
The sources for this piece include an article in NPR.
