Many users of the intelligent ChatGPT platform have expressed their dissatisfaction with the service problems, and showed it through their social media accounts. These problems are that they are unable to log in and the processing of requests is very slow.
This problem is manifested in a major outage in the chatbot services that other websites, platforms and services use to provide their services to the public.
The inconvenience caused to subscribers of the paid service comes from ChatPGT, the famous platform for artificial intelligence, who objected to their inability to access the service or its very slow delivery, and it continues to this moment.
Doan Detector, which specializes in monitoring services and websites, confirms an increase in the number of complaints being filed due to ChatGPT outages.
It should be noted that this error started in the early hours of Monday morning, when users from all over the world began sharing screenshots and images showing the service not working and their inability to access the site.
OpenAI reported that a problem occurred in the company's public chat unit, and the problem was resolved, but the problem still exists as of the date of writing this news.
OpenAI has unveiled a new paradigm for artificial intelligence called GPT-4, which is a significant evolution of the previous model, GPT-3.5.
Paragraph alternative in Arabic: The latest generation of GPT-4 is classified as "multi-model". This generation is distinguished by the availability of the feature of entering images in addition to the texts in it, in contrast to previous generations of it that could only enter the text. GPT-4 can handle the content inside images, and perform the analysis and explanation of their content. It was revealed, for example, that this alternative allows extracting information about the material reviewed in an easy way, as it is not algebraic like the traditional methods of thinking. A new study includes images and graphs, and will enable GPT-4 to read and analyze them, answer questions about them, and then explain it in the text report.
According to the company, GPT-4 showed results approximating human performance after being tested in a range of professional and academic tests. This new model passed a comprehensive test that simulates the practice exam with a rating that puts it in the top ten percent of the testers, while the GPT-3.5 in the same exam placed it in the bottom ten percent.
The company reported that they have been working on improving GPT-4 over the past six months, based on their experience with ChatGPT, and they also noted their partnership with Microsoft in developing a new computer from scratch to create the infrastructure required to run this new model, which requires a great deal of capital. processing power.
In the end, the company reported that its new model has improved significantly, but is still as susceptible to AI errors as other models. It advised users to take care and caution in using this application, and to ensure the authenticity of the content before adopting it, and also warned against using it in tasks that require high accuracy and proficiency.