OpenAI, a Microsoft-backed startup, will not be training a successor to GPT-4 “for the time being” due to concerns about the speed of GPT-4’s progress.
The American startup launched ChatGPT last November, powered by the GPT-3 large-scale language model.
OpenAI released GPT-4, one of the most anticipated generations of LLM, in March. GPT-4 is the latest and most advanced version of the company’s large language model, which is embedded in ChatGPT and several other applications. But fans and commentators are already eyeing GPT-5.
just last week At a conference held by the Government of India economic times The OpenAI CEO was asked when he plans to start training the next LLM.
“We have a lot of work to do before we start that model. rice field.
Following the announcement of GPT-4 and other related developments in the era of generative AI, over 1,000 technology leaders wrote an open letter to help developers better understand how these technologies work. We called for a pause in all major AI development and training until
Elon Musk, Apple co-founder Steve Wozniak and other prominent figures signed the letter, which gathered more than 1,377 signatures.
The letter included Turing Award winner Joshua Bengio, computer science professor Stuart Russell, Stability AI CEO Emad Mostak, Getty Images CEO Craig Peters, and several other tech executives. It also contained the signature of a prominent scientist.
A few weeks later, Altman said the letter was “mostly devoid of technical nuance as to where a pause might be needed,” but the intention was that OpenAI would be training GPT-5. It had not started and had not been in immediate plans “for a while”.
Interestingly, @Open AI not yet trained #GPT5successor to #GPT4 it’s behind #chatGPT Due to concerns over the dangers posed by misuse of large language models (#LLM|s). Click here for details: https://t.co/sFYZ353lIo
— Oliver Lord (@olivertlord) June 9, 2023
Altman reiterated concerns raised by prominent voices in the AI space. Altman said the company is taking proactive steps to mitigate potential risks associated with AI by implementing strict security measures.
“When we completed GPT-4, it took over six months to be ready for release,” says Altman.
Altman has previously emphasized in an interview that OpenAI opposes regulating small AI startups.
“The only regulation we ask is for ourselves and the larger public,” the CEO insisted.
Also read: White House Takes Steps To Study AI Risks And Assess Impact On Workers
Waiting for a better GPU?
There is speculation on Twitter that OpenAI is waiting for better hardware to train the next generation of powerful GPT models.
“Waiting for better data from GPU and GPT-4 interactions.” tweeted Bert Kastel’s reaction to the news about the OpenAI delay.
“We don’t have enough hardware for GPT-4 either,” argues Christoph C. Cemper, AIPRM’s Chief Prompt Officer.
Not enough hardware for GPT-4
— Christoph C. Semper 🇺🇦 🧡 SEO&AI (@cemper) June 8, 2023
Oliver Lord, Chair of the Department of Petrography at Bristol University of Geosciences, said: mentioned OpenAI has yet to start training GPT-5, the successor to GPT-4, the model that powers ChatGPT.
The reason behind this decision, Lord argued, was concern about the potential dangers associated with misuse of LLM.
Microsoft to offer GPT model to governments
Technology giant Microsoft is using Azure cloud services to bring OpenAI’s language generation models to US federal agencies. according to to Reuters.
The company has expanded its support within Azure Government to include integration of OpenAI’s advanced LLMs, including the latest and most sophisticated GPT-4 and GPT-3.
This is Microsoft’s first introduction of GPT technology into Azure Government, which provides cloud solutions to US government agencies.
This is the first effort by a major company to make chatbot technology available to the government.