Artificial intelligence GPT3 writes texts independently, even passing legal exams – science is in disarray. A 22 year old’s app now needs to display GPT3 texts.
Edward Tian studies computer science and journalism at Princeton University and has experienced firsthand the debate about artificial intelligence at the nonprofit OpenAI. Their latest project, ChatGPT3, is able to independently write articles, term papers and entire scientific papers that are virtually indistinguishable from those of human authors.
As a result, universities around the world are alarmed – New York educational institutions have completely banned the use of ChatGPT3. Edward followed the discussion into the semester break.
22-year-old AI programs that recognize AI texts
There is increasing uncertainty on the Internet as to whether a user is truly a human being. It is not uncommon for social media platforms or websites to fall victim to attacks by so-called bots that can flood servers with information and thus paralyze them.
Therefore, operators require potential users to participate in reverse Turing tests, which are now standard for PayPal transactions, authentication for ridesharing apps, or the creation of social media accounts. Here the user is given a calculation or assignment task that is easy for humans to solve and bots fail.
There was no such test before for artificial intelligence like GPT3. Edward Tian, 22, was troubled by this problem. In the basement of his family’s house in Toronto on New Year’s Eve, he programmed an application that was said to be able to identify texts written by ChatGPT3. GPTZero was born.
ChatGPT will be beaten at its own game
The Princeton student used artificial intelligence software as the basis for his practice, which is a headache for professors all over the world: GPTZero needs to defeat ChatGPT3 with its own weapons, artificial intelligence needs to unmask. It seems that a simple but ingenious solution has been found against the abuse of GPT3.
Since the launch of GPTZero in early January, Edward Tian has caught the attention of investors, professors and journalists around the world. The app is currently free and generally accessible from the internet.
Critics: GPT3 will learn to go unnoticed
Criticism is coming from experts, however: Toby Walsh, senior professor of artificial intelligence at the University of South Wales, thinks GPTZero is doomed to fail. In the long run, the app works against its plan to be able to easily identify texts written by AI.
Why: Because GPTZero uses the same publicly available OpenAI software as ChatGPT3, artificial intelligence will learn best to stay undetected. The reactions of GPTZero users are also quite mixed: the software often does not work or gives incorrect results.
Inventor wants to make AI use safer
Tian is aware of this criticism, but sees GPTZero still in its testing phase and hopes investors can address these teething issues. The 22-year-old player doesn’t think much of an artificial intelligence ban like ChatGPT3 anyway.
In an interview with US broadcast syndicate NPR, he said GPTZero “should not be a tool to hinder the use of artificial intelligence. But as with any new technology, we must be able to use it responsibly and take precautions against abuse.”
I’m Ashley Robinson, a professional writer and journalist. I specialize in news writing and have been working for the past five years with News Unrolled. My main focus is on technology-related topics, though I also write about politics, healthcare, and business from time to time.