In the past few years, ChatGPT has taken the Internet by storm. Artificial intelligence has not only caught the attention of the business community, but it has also spawned a host of similar service providers as well. As we have shared with you recently, Microsoft has invested $10 billion in OpenAI, and Baidu is working on an AI system that is a competitor to OpenAI. However, not everyone is as happy with ChatGPT as Microsoft is with it. After just two months on the internet, there was a great deal of AI-generated content that was flooding the web with new content. Wanting to fix the issue, OpenAI releases tool to detect AI-generated text, including from ChatGPT. You can find more details about the event here…
ChatGPT and AI-Generated Content Under the Microscope with OpenAI’s Latest Launch
The OpenAI team has just released a tool that allows you to distinguish between text written by a human and that of an artificial intelligence, after months of debate and confusion. Despite the fact that this software, which detects both the company’s own ChatGPT and other artificial intelligence tools, is a good start, it is not enough to effectively combat this kind of threat. In my opinion, the success rate of this tool called “AI Text Classifier” is relatively low, at around 26%, which is not good. This problem has also been stated on the website for OpenAI, which will most likely improve the system in the future as a result of this.
There is no doubt that we were eager to give the tool in question a try, so we prepared two texts to test it. In one of them, ChatGPT has written a paragraph and in the other, I have written a paragraph, who is also the author of this article. There was a bit of a disappointment with the results, as we weren’t quite sure what to expect. A warning was issued to both articles informing them that the classifier deems the text to be unclear if the text was generated by AI.
READ MORE: In order to prevent account sharing outside of the household, Netflix lists rules and exemptions
It is important to note that although OpenAI acknowledges the margin of error, it also warns its users about the issue. According to the statement that can be found on the company’s website, the AI Text Classifier is intended to foster conversations about the difference between human-written content and AI-generated content by sparking a discussion about the distinction between the two. The results of the analysis may be deemed useful in determining whether a document was generated with the aid of AI, but they should not be the only evidence when doing so. The model is trained on human-written text from a variety of sources, which may not be representative of all kinds of human-written text.”
“AI Text Classifier” also has some limits. Current limitations:
- Requires a minimum of 1,000 characters, which is approximately 150 – 250 words.
- The classifier isn’t always accurate; it can mislabel both AI-generated and human-written text.
- AI-generated text can be edited easily to evade the classifier.
- The classifier is likely to get things wrong on text written by children and on text not in English, because it was primarily trained on English content written by adults.