OpenAI terminates tool for detecting text written by AI “bad accuracy”

The observant may remember the release of the OpenAI AI text classifier in early 2023.Well, it’s no secret that the latter doesn’t represent the breakthrough that some initially thought, but it wasn’t necessarily everyone’s bet. tool closure.

But a quick addition about the overall initial announcement post popped up on the official OpenAI portal. In the latter, which is only now marked in a significant way, it says from July 20, 2023. AI Text Classifier is no longer accessible. Whether or not it’s an issue that will be “solved” in the short term is unclear, but the decision is certainly not going to be ignored.

This is also the reason why OpenAI itself cites Survey “poor accuracy” by tools. The organization says it is already working on more accurate methods of using artificial intelligence to detect what text is actually written, as well as implementing a feedback system.

In addition, OpenAI is working on creating tools for understanding whether audio or visual content was also created by AI. On the other hand, remember that your organization is one of their realities. The Biden administration promised to take action More specific in the area of ​​security. To be clear, we are referring to the rise of watermarks in the AI ​​space, so the timing of OpenAI’s decision to “put aside” AI text classifiers does not seem to be a coincidence.