Much has been said about artificial intelligence and especially about ChatGPT. The intelligent system relies on GPT-3.5 language technology and its purpose is to optimize the language models intended for intelligent dialogue. On the other hand, there are those who are already thinking about new versions of this service, but the creator OpenAI made it clear that it does not train GPT-5 and that this system will not arrive soon.
GPT-5 is not trained, OpenAI guarantees
It was during a lecture that took place last Thursday (13) that Sam Altman, CEO of OpenAI declare that the company is not training GPT-5, its artificial intelligence language model (LLM), and that it will not arrive anytime soon.
On the other hand, even if the development of GPT-5 does not take place, Altman said that the company is focused on improving GPT-4, its most recent language model released on March 14, 2023, which is used on ChatGPT Plus and Microsoft's Bing Chat.
The lecture where these statements took place is called "The Future of Business AIs", an event organized by the Massachusetts Institute of Technology (MIT).
In this way, the executive finally denied the rumors surrounding this version and the concerns about it, which gave rise to a letter asking for the suspension of the development of AI services of the kind, signed by names such as Elon Musk , Steve Wozniak , among others. But Altman also adds that this letter was not explicit and that it did not say exactly what they wanted the companies to stop and how they should do it.
For the CEO, the short- and medium-term goal is not to train the GPT-5, but rather to improve the system he already has in his hands, since the GPT-4 still has some bugs, even in simple tasks.