As we all know, the ChatGPT is an AI Chatbot by the OpenAI. It is very widely used now a days, specially the free version they provide. Previously, the ChatGPT had a feature, where we were able to feed it some text and the chatbot was able to recognize the text and tell whether text was written by ChatGPT itself or not. This feature of ChatGPT was called OpenAI Classifier.
However, OpenAI have decided to discontinue and remove the feature from ChatGPT. In fact, it has already been removed. Now the ChatGPT cannot tell if the text was written using ChatGPT, another AI tool or a human. According to OpenAI, the Classifier is removed due to its low level of accuracy. The results it returned were not fully accurate and ChatGPT was struggling to differentiate between the text written by itself or humans or other chatbots.
It is great that ChatGPT took this step in timely manner. As some teachers were using this built in classifier tool along with other third party tools to see if the certain assignment is written by student himself/herself. One wrong classification of an assignment can have catastrophic consequences for the student and his/her career.
However, still there are some third party tools available which detect that certain text is written by an AI (Artificial Intelligence) tool or not. They also provide a confidence score with the evaluation, i.e. how confident the tool is, that this text was indeed written by an AI tool.
However, still one misevaluation can have very bad effects on careers or atleast grades of the students. So relying fully on such tools is not the solution now a days. Teachers can adopt other methods for scrutiny, like asking questions from students about the text written by them in the assignments. Which means, if student have even used AI tools to write the assignments, the teachers does not necessarily need to use them to detect if the assignment was written using an AI tool.
AI tools like ChatGPT by OpenAI, Bard by Google and more, are getting very intelligent day by day. So they are also getting better at writing the texts more naturally like humans. So any AI detection tool which still works, there is a chance that it will stop working or giving the inaccurate results in near future. So it is high time that teachers and anyone else who wants to detect the AI generated text, start using alternate methods, like the ones mentioned above.