ChatGPT: no longer possible to cheat with AI, this student has developed the perfect counterattack


While teachers around the world are wondering how they will be able to fight against the ChatGPT wave among students, developer Edward Tian has just brought them the solution on a set. GPTZero is an application capable of detecting, using a simple copy-paste, texts generated by artificial intelligence.

Credits: phonlamaiphoto/132rf

This is an application that should appeal to all teachers who have banned ChatGPT from their courses. Edward Tian, ​​a computer science and journalism student at Princeton University, spent a rather special New Year, taking advantage of his few days off for GPTZero. You will have understood from its name, this web-application is a direct response to ChatGPT, the AI ​​that has been driving Internet users crazy since December.

The promise is enticing. Edward Tian claims that GPTZero is able to detect any text generated by an artificial intelligence. To do this, simply copy and paste the text in question on its site,, and let the application do its job. In order to determine the “true” from the “false”, or rather, the human from the machine, the algorithm will calculate the complexity rate of the submitted text.

Here is the perfect tool to spot texts generated by ChatGPT

For this, GTZero also analyzes the randomness of the text by comparing it to a similar pattern. The application then takes care of evaluating the overall consistency, the last link in the chain which allows it, in theory, to detect the AI ​​which is possibly hiding behind the text. And, according to the videos posted on Twitter by its creator, it is clear that the tool is very effective.

On the same subject — Bing: Microsoft will boost its search engine with ChatGPT to compete with Google

Edward Tian has indeed tested its application on an article written by journalist John McPhee, then a LinkedIn post generated by ChatGPT. In both, GPTZero hit the nail on the head, and even has the luxury of displaying its results in the form of a graph, to identify the details that spilled the beans. If you were hoping to write your next philosophy dissertation with the chatbot, sorry, but it’s already too late.


Source link -101