OpenAI said about the start of training for a new flagship model of artificial intelligence. It will replace GPT-4, which manages the ChatGPT chatbot. The company declares the next level of capabilities, and in particular, the desire to create no less than general artificial intelligence (AGI) — AI that can do everything the human brain can. The new AI model will drive chatbots, digital assistants, image generators, and search platforms.
«OpenAI has recently begun training its next advanced model, and we expect the resulting systems to take us to the next level of capability on our journey to AGI. While we pride ourselves on building and releasing models that are industry-leading in both capability and security, we welcome robust discussion at this important time» — the company blog says.
The company also announced that it is establishing a «Safety and Security Committee (SSC) to investigate the risks associated with the new technology.The «Committee» includes Sam Altman, as well as board members Bret Taylor (chairman of the OpenAI board and co-founder of the customer service startup Sierra AI), Adam D’Angelo (CEO of Quora and the artificial intelligence model aggregator Poe), and Nicole Seligman (former executive vice president and general counsel of Sony Corporation). The company said that the new rules it has developed could be put into effect in late summer or fall.
Experts do not have a consensus on when general artificial intelligence will be created, but companies are steadily increasing the power of the technology, demonstrating a noticeable leap every two to three years. Training of AI models can take months or even years. After the training is complete, companies usually spend a few more months testing the technology and fine-tuning it for general use. This may mean that the next OpenAI model will only appear in nine months, a year, or more.
Sources: The New York Times, Venture Beat