OpenAI Shuts Down Its AI Detector Due To Its Poor Accuracy

Due to the low accuracy of its AI Classifier, OpenAI has discreetly discontinued it.

Source: appmaster.io

Powerhouse in artificial intelligence, OpenAI has covertly stopped developing its AI-detection program, alleging a poor accuracy rate. The OpenAI-created AI classifier was initially released on January 31 with the goal of helping users, including instructors and academics, tell human-written content from AI-generated text.

However, the AI decoder has been discontinued as of July 20 according to the initial blog post that announced the tool's launch. The AI classifier is no longer available as of July 20, 2023, due to its low degree of accuracy. The message simply provided a brief justification for why the tool had been taken down, and the link to it is no longer active. The business did clarify, though, that it was investigating fresh, better ways of detecting AI-generated material.

The statement said, “We are dedicated to creating and implementing technologies that enable users to recognize if audio or visual material is produced by AI. We are working to incorporate criticism and are now investigating more efficient provenance techniques for text.”

OpenAI was upfront about the fact that the identification tool was prone to mistakes and could not be regarded as fully reliable. The business said that the tool's shortcomings included the ability to confidently classify material produced by people as AI-generated and being very inaccurate at authenticating text with fewer than 1,000 characters.

Apple Has Developed Its Own GPT AI System, But There Are Currently No Plans To Offer It To The Public

In a study released on July 18, experts from Stanford and UC Berkeley found that ChatGPT, the company's main product, was becoming noticeably poorer over time. Researchers discovered that ChatGPT-4's capacity to correctly identify prime numbers has dropped from 97.6% to only 2.4% during the previous few months. Additionally, the ability to produce new lines of code significantly decreased for ChatGPT-3.5 and ChatGPT-4.

Apple, situated in Cupertino, California, is one of the most significant participants in AI, even if it hasn't had much of an influence on the chatbot market. The AI that powers the iPhone's camera and photo editing software is still state-of-the-art, and Apple Research consistently publishes important papers in the field of machine learning. The GANfather, Ian Goodfellow, who lately left the organization to join Google DeepMind, as well as its present head of AI, John Giannandrea, who before oversaw Search at Google, have also recently made their way through the company's secretive AI labs.

Disclaimer: FameEX makes no representations on the accuracy or suitability of any official statements made by the exchange regarding the data in this area or any related financial advice.