Thank you for reading this post, don't forget to subscribe!
ChatGPT creator OpenAI has now unveiled a tool called ZeroGPT, with which, it says, users will be able to detect whether text has been written by an artificial intelligence (AI) programme and passed off as human-written, or it was actually written by a person.
How does ZeroGPT work?
According to OpenAI, users need to enter the text in the box and click on ‘Detect Text.’ Using algorithms developed by the ZeroGPT team, the AI software will check the input and publish the result.
What do the results show?
ZeroGPT will show one of the following results:
(1.) The text is human-written
(2.) The text is AI/GPT generated
(3.) Most of it is AI/GPT generated
(4.) It is most likely AI/GPT generated
(5.) It is likely generated by AI/GPT
(6.) The text is ‘mixed,’ with some parts generated by AI/GPT
(7.) The text is most likely human-written
(8.) It is most likely human written, and includes parts generated by AI/GPT
How accurate is ZeroGPT?
ZeroGPT uses the DeepaAnalyse technology to identify the origin of the input. Currently, its accuracy rate stands at more than 98%, but OpenAI says it is trying to bring down the error rate to less than 1%.
How was ZeroGPT developed?
For this, the San Francisco-based research lab says it analysed more than 10 million articles – those generated by AI, others human-written – to develop algorithm for ZeroGPT. Experiments, however, are still ongoing.
Anything else?
The tool will mainly benefit students, teachers, educators, writers, freelancers and copywriters. Also, it detects the AI text output in all the available languages.
