Having promised to suspend work on the successor to the GPT-4 neural network, OpenAI never started training GPT-5. The measure is prompted by public concern about the potential threat of large language models to the well-being of humanity.
Delay in development of GPT-5
At a conference hosted by the Indian newspaper Economic Times, OpenAI CEO Sam Altman revealed that the company is not even close to launching the GPT-5 model. More work needs to be done and new ideas developed to keep the system safe and ethical.
Open letter and calls for suspension
More than a thousand members of the public and the expert community, including well-known figures such as Elon Musk and Steve Wozniak, signed an open letter calling for the suspension of training on systems more powerful than GPT-4 for a minimum of 6 months. However, Altman noted that the paper did not provide enough technical detail on exactly where to pause, and confirmed that GPT-5 training has not been started or planned for anytime soon.
AI Security
The head of OpenAI dismissed some concerns about potential threats to artificial intelligence systems. He stated that OpenAI has the ability to conduct a comprehensive security assessment with external experts, simulate full-scale cyber attacks, and perform other security tests. Altman stressed that the development of GPT-4 already required six months of work after completion before the platform was released to the public.
Against government regulation
The head of OpenAI opposed the proposed government regulation measures for small start-ups in the field of artificial intelligence, notes NIX Solutions. Instead, he proposed holding Senate hearings and regulating OpenAI itself and other major market participants in order to ensure the safety and ethics in the development and use of artificial intelligence.