ChatGPT 4, the popular language model developed by OpenAI, is facing challenges with accuracy and speed, according to a recent study by Stanford University. The study reveals that ChatGPT 4, which powers the ChatGPT app, has become slower and less proficient at math over time.
The app, which is set to be released soon on Android devices, is already available on the Apple App Store. It positions itself as a virtual assistant capable of providing instant answers, tailored advice, creative inspiration, professional input, and learning opportunities. Users will need an OpenAI account to access the app, just like the web version and the iPhone app.
Despite the considerable excitement surrounding ChatGPT, the app currently ranks at number 16 on the list of most popular free apps on the Apple App Store. It falls behind apps like Temu, Threads, Vinted, and Shein, leaving one to wonder whether everyday AI can compete with the allure of fast fashion.
The Stanford University study highlights the fluctuating performance of large language models like GPT-3.5 and GPT-4. In specific tests, GPT-4’s accuracy dropped from an impressive 97.6% in March to a mere 2.4% in June. The study emphasizes the need for consistent monitoring of AI models to ensure their performance remains reliable.
OpenAI recently discontinued its AI Classifier tool, which aimed to identify text generated by AI. The decision to discontinue the tool was due to its low accuracy rate. However, OpenAI assures users that it is actively working on developing more effective techniques for tracing the origins of AI-generated content.
Meanwhile, OpenAI’s CEO, Sam Altman, has shifted his focus to a new venture called Worldcoin. Altman, together with Tools for Humanity CEO Alex Blania, has launched Worldcoin—a cryptocurrency that utilizes an eyeball-scanning Orb to verify a person’s identity while preserving their anonymity.
In light of these developments, users and businesses relying on AI technologies in their workflow are advised to remain vigilant and monitor performance to avoid any decline in accuracy or speed. OpenAI acknowledges the importance of consistently improving and refining AI models to deliver reliable and efficient results.
As the ChatGPT app prepares to make its debut on Android, it faces the challenges posed by its competitors on the Apple App Store and the scrutiny brought by the Stanford University study. How everyday AI can compete with fast fashion remains an open question, but ongoing advancements and refinements in AI models offer hope for continued progress in the field.
