“It is uncertain whether the text was written by artificial intelligence (AI).” The AI Classifier, which is programmed to discern whether a text is written by an AI or a human, repeatedly answered that it is uncertain. This is an AI program that OPEN AI, a developer of text-generating AI, unveiled Monday (local time). The classifier is trained to compare texts written by humans and texts written by AI.
This AI was developed to calm the increasing controversies over the side effects of ChatGPT - as workers and students have been using the program for their assignments. The US NBC broadcast reported, “We are paying attention to whether it can become an AI that catches AI.”
This is how to use the AI classifier. After pasting the text that needs to be determined in the blank space of the classifier’s website and submit it, the results come out soon after whether the possibility of the text’s author would AI as in a scale of “very high,” “uncertain” and “unlikely.”
So far, many say the result is unsatisfactory. When a 1,700-letter summary of the British novel “Pride and Prejudice” produced by ChatGPT, was submitted, the outcome was “uncertain,” meaning that the AI is unsure whether the text is from AI. It does not help that the program can only decide on texts in English. OpenAI mentioned that “internal text shows 26% of accuracy. 9% of human writing was determined as AI wrote,” but said the accuracy could advance if the AI goes under more training.
Hyoun-Soo Kim firstname.lastname@example.org