According to reports, a Princeton University student in the United States has created a tool that can tell when a student has used an AI chatbot to create an essay rather than writing it themselves.
The internet has been astonished by ChatGPT’s ability to write lines and lines of convincing written responses based on basic inputs provided by human users ever since it became publicly accessible at the end of 2022.
As a result, many people have pondered how education will develop in the future and how kids might ultimately use technology and artificially intelligent software to complete all of their schoolwork for them.
But now, Princeton’s own Edward Tian argues that his programme, called GPTZero, can identify the incidence of such behaviour, even if it might have cost him some peer favour.
The 22-year-old Computer Science student wrote on Twitter, “I spent New Years inventing GPTZero – an app that can quickly and effectively detect whether an essay is ChatGPT or human-written.
He went on to display a demo of the programme in action, classifying a New Yorker essay by John McPhee as having been authored by a person while classifying a LinkedIn social post as having been written by a bot.
The “perplexity” feature of the app is said to determine how random the sentences are and whether or not the app thinks the constructed sentence is odd.
Then it proceeds to assess the “burstiness” of the content, which is done by assessing how similar the sentences are to one another. There is typically more variety in sentences that are written by humans.
Following his Twitter announcement of GPTZero, Edward published a Substack article claiming that the software had been downloaded by over 10,000 users. It had been tried out by so many people at one point that the website had malfunctions.
The possibility of widespread “plagiarism” utilising AI systems like ChatGPT, according to Edward, was the driving force for him developing the software during his free time.
“There is a lot of excitement about ChatGPT. Has AI written this and that? Humans have a right to know! “said he.
A programme like GPTZero, for example, could undoubtedly be in high demand, given the fact that millions of people have downloaded it to test out its features and that many others use it as a tool to complete tasks for them with no effort.
There have been recent speculations that even Microsoft is considering integrating ChatGPT capability into their suite of Office productivity tools, which further raises the potential that kids may try to use AI to complete their homework for them in an effort to get by in school.
This simply raises the intriguing question of whether or whether AI platforms could eventually create essays that were intricate and challenging enough to deceive detection tools like GPTZero, which could then need to be updated frequently to be effective in spotting essays that were produced by AI.
