OpenAI, San Francisco, CA-based leading artificial intelligence research institute, announced today the launch of a new AI classifier designed to distinguish between text written by a human and text generated by AI systems.
The classifier was trained on a diverse range of AI-generated text and has shown promising results in detecting such text. In evaluations on a “challenge set” of English texts, the classifier was able to correctly identify 26% of AI-generated text as “likely AI-written” while only incorrectly labeling human-written text as AI-written 9% of the time. The classifier’s reliability improves with longer input texts and has shown significant improvement compared to OpenAI’s previous classifier on text from more recent AI systems.
OpenAI’s new AI classifier has the potential to mitigate false claims that AI-generated text was written by a human, such as in the case of automated misinformation campaigns or the use of AI tools for academic dishonesty. The classifier can also help detect AI chatbots posing as humans.
The classifier is being made publicly available to receive feedback on its usefulness and OpenAI plans to continue its work on the detection of AI-generated text with the hope of sharing improved methods in the future.
“We’re proud to be at the forefront of AI research and development, and our new AI classifier is just one of the many exciting projects we have in the works,” said OpenAI’s CEO, Sam Altman. “Our goal is to use AI for the betterment of society and this new classifier is just one step in that direction.”
OpenAI’s new AI classifier is an important development in the field of AI and will likely have far-reaching implications for businesses, governments, and individuals alike. The company’s commitment to transparency and ethical AI practices is evident in the public release of this classifier, and the world can expect to see more groundbreaking developments from OpenAI in the future.
