ChatGPT Creators Launch New Tool That Can Detect AI-Written Articles
Create demand, provide supply. :P
OpenAI, commonly known to be the brains behind the revolutionary AI-writing platform ChatGPT, has come up with a new AI classifier that detects whether an article was written by artificial intelligence
The AI Classifier is aimed at distinguishing between text written by humans and text written by AI. This move was done in order to combat issues such as such as students using AI chatbots to write for assignments.
The results of the AI Classifier are labeled as: Very unlikely, unlikely, unclear if it is, possibly, or likely AI-generated.
As of the time of writing, OpenAI said that their AI Classifier still faces certain limitations
According to test runs, their classifier correctly identified 26% of AI-written text (true positives) as "likely AI-written," while incorrectly labelling human-written text as AI-written 9% of the time (false positives).
However, they noted that the reliability of the platform increases with the more text you put in.
Naturally, we had to try the classifier for ourselves
It's perfect because we recently wrote an entire piece about ChatGPT using ChatGPT.
After inputting our text in the classifier prompt, it considered our text to be 'likely generated by AI'.
So it seems like it knows what it's doing. :P