The AI chatbot ChatGPT developed by OpenAI is very popular among Internet users looking for answers to various questions and tasks in seconds.
However, this trend has caused serious concern among educators who worry about students submitting work written by ChatGPT.
In response, a Stanford University researcher developed his tool, called DetectGPT, to help teachers identify content generated by her ChatGPT and other similar large-scale language models (LLMs). bottom.
After New York State banned public school students from using his ChatGPT for academic purposes, OpenAI confirmed it was in the process of creating a detection tool.
DetectGPT is a tool created to identify text written by ChatGPT or similar chatbots. The researchers behind it say that these AI-generated texts are “evidence of the negative region of curvature of the model’s log-likelihood function.”
This information helps DetectGPT determine whether the text was written by a particular language model by checking log probabilities. Researchers claim that DetectGPT has him right 95% of the time. The logarithmic probability function is a mathematical concept used in the fields of AI and machine learning. Used to determine the probability or confidence level that an event will occur.
Logarithmic transformation of probabilities facilitates comparison and interpretation of results, especially in applications such as natural language processing and language generation. This allows for more accurate evaluation of texts generated by AI models. Educational institutions are concerned that their students are using tools like ChatGPT to write assignments and homework, so they developed a tool called DetectGPT to detect them.
Unfortunately, the team behind DetectGPT has yet to reveal much information about the tool and it is currently closed to the public.
Still, developing such a tool is a step in the right direction to ensure that students do their work in an honest and authoritative manner.