The creator of ChatGPT created a free tool to detect AI-generated text
OpenAI; the company behind DALL-E and ChatGPT has released a free tool that it says is intended to "distinguish between human-written text and AI-written text." He warns that the classifier is "not entirely reliable" in a press release and "should not be used as a primary decision-making tool." According to OpenAI; this can be useful in trying to determine if someone is trying to pass off the generated text as something that was written by a person.
The tool ( known as the Classifier) is relatively simple; but you'll need a free OpenAI account to use it. You simply paste some text into a box; click a button and it will tell you if it thinks the text is highly unlikely; unlikely; unclear if it is; possibly; or probably AI generated.
OpenAI Launches GPT-Classifier to Help Check AI-Written Text
In testing; it identified 26% of AI-written text as "probably AI-written"
The tool uses natural language processing algorithms and machine learning models to analyze conversations for signs of AI-generated content. It then provides users with information on whether their conversation partner is likely an AI system or not. The results also include details about ChatGPT's confidence in its analysis so users can make more informed decisions about pursuing interactions with potential online bots.
The new GPT-Classifier attempts to determine whether a given text was written by a human or the work of an AI generator. While ChatGPT and other GPT models are largely trained on all kinds of text input; the GPT-Classifier tool is "honed on a data set of human-written and human-written text pairs. IA on the same topic".
GPT-Classifier detection is currently limited
While the idea of easily spotting AI-generated text will be music to editors and educators' ears; OpenAI warned that its "classifier is not entirely reliable."
Currently; GPT-Classifier "correctly identifies 26% of AI-written text" while labeling 9% of human text as AI-written. OpenAI also notes that the tool's accuracy "generally improves as the length of input text increases."
The main limitations of GPT-Classifier are its cost, sample efficiency during pre-training, and its inability to understand language [1][2][3][4][5].
20 AI Content Detectors
1. Natural language processing (NLP)
2. Sentiment analysis
3. Text classification
4. Parts-of-speech tagging
5. Named entity recognition
6. Text summarization
7. Machine translation
8. Text clustering
9. Text generation
10. Text analytics
11. Document summarization
12. Automated essay scoring
13. Text similarity detection
14. Text-based search
15. Text-to-speech synthesis
16. Chatbot development
17. Textual entailment
18. Automatic summarization
19. Grammar correction
20. Automatic text correction