Open AI has recently released a language model classifier that aims to identify text written by humans from text generated by AI. The aim is to combat the spreading of misinformation that claims AI-generated text is written by humans.
- Classifier has 26% accuracy in identifying AI-generated text, 9% false positive rate for human-written text
- Classifier struggles with short texts, non-English languages, and predictable text
- Open AI recommends using the classifier as a complement to other methods
- Classifier trained on dataset of human-written and AI-written text on the same topic
- Open AI develops preliminary resource for educators on ChatGPT language model and seeks feedback
- Represents a step forward in combating misinformation and misuse of AI-generated text.
