GPT-4 ensures that ChatGPT can talk to us like a human. A new model that could exceed its capabilities.
OpenAI announced that it is working on… Successor to GPT-4 He works. It can be read in one go Blog post is actually about creating a safety committee He goes, As the New York Times discovered.
The entry states: “OpenAI recently began training its next model and we expect the resulting systems to take us to the next level of performance.” The successor to GPT-4 could be used in the future for chatbots like ChatGPT, but also for Digital assistants Like Apple's Siri, search engines And image generators.
➤ Read more: Scarlett Johansson argues with OpenAI: AI's voice sounds like hers
Development will take a few more months
The current version of ChatGPT, GPT-4, was released in March 2023. OpenAI recently released a slightly improved version of it GPT-4o Baptized. AI models learn automatically by… Analyzing large amounts of data. This can include texts from the Internet and books, but also movies, music and other data.
Training such models takes between a few months and several years. Once the training phase is complete, AI companies usually test their models for a few months until they are made available to the general public. And so it should At least 9 more months Wait for the new model to come out.
➤ Read more: Neural networks: how artificial intelligence is constantly learning
OpenAI's AI security remains a sensitive issue
According to New York Times journalist Kid Metz, OpenAI will try to be faster than its competitors in the field of artificial intelligence. Opposing voices However, they warn that developments in the AI sector are happening too quickly and the technology is gradually becoming so A danger to humanity They will because they are people Careers Take away and misinformation It can spread.
According to the blog post, the new security committee will investigate how best to deal with AI risks. There have been similar attempts at OpenAI in the past, but their success is questionable.
Recently, the former security chief and co-founder of OpenAI left Ilya Sutskever, the company. He felt that AI companies in general did not care enough about AI security. Last fall, Sutskever played an integral role in this OpenAI CEO Sam Altman fired From the Supervisory Board – In his opinion, Altman did not care enough about the security of AI at OpenAI. Sutskever left the company in May, while Sam Altman was brought back to the company shortly after his dismissal last year.
➤ Read more: Disruption in OpenAI: Behind the Chaos in ChatGPT Maker
