Dimitrios Kambouris/Getty Images for DVF
- OpenAI's chief tech officer wants less hype around GPT-4, an update to the model that powers ChatGPT.
- MIra Murati's comment comes as rumors around GPT-4's release and capabilities spread.
- OpenAI has not said when it will release GPT-4, though Murati said it may "broaden opportunities."
Many in the tech world are anxiously waiting OpenAI's release of GPT-4, the more powerful iteration of the technology that runs ChatGPT. But Mira Murati, OpenAI's chief technology officer, wants the world to chill out a little.
"I think less hype would be good," Murati told Fast Company when asked about GPT-4, an upgrade to GPT-3.5, the language processing model that currently runs ChatGPT.
Murati's comment comes as rumors continue to spread over GPT-4's release date and capabilities.
AI experts and Silicon Valley types say that GPT-4 will be a major advancement, as they believe it will be trained on hundreds of billions of more parameters — or data points — than GPT-3, The Verge reported. More parameters would allow the AI to scrape more data from the internet and, theoretically, make it more accurate.
Some even believe that GPT-4 may lead to artificial general intelligence — the ability for AI to perform complex human tasks like consciousness, per The Verge. But OpenAI's CEO Sam Altman shut that rumor down.
When he was asked in a StrictlyVC interview about a chart comparing the billions of parameters of GPT-3 with the alleged trillions of GPT-4, Altman said it was "complete bullshit."
"The GPT-4 rumor mill is a ridiculous thing," Altman said in the interview with StrictlyVC. "People are begging to be disappointed, and they will be."
There were rumors that Microsoft's new Bing search engine would run on GPT-4, though those were quickly debunked after Microsoft announced that it uses a GPT-3.5 model — a fine-tuned version of the GPT-3 model — during its launch.
Despite speculation around the elusive model, Murat did tell Fast Company that GPT-4 may be able to "broaden opportunities for people."
Referring to the the controversy around ChatGPT's usage in schools, Murat thinks that GPT-4 will turn disgruntled teachers who believe ChatGPT encourages cheating and plagiarism into happy teachers who can use the chatbot as a tool to, say, help write lesson plans.
"With ChatGPT, you can have this infinite interaction and have it teach you about complex topics in a way that's based on your context," Murati told Fast Company. "It's kind of like a personal tutor."
As of now, GPT-4 remains under wraps, and CEO Altman has not announced when he will release it.
"It'll come out at some point, when we are confident we can do it safely and responsibly," he told StrictlyVC.
Do you work in generative AI or have insight to share? Reach out to the reporter from a non-work email at amok@insider.com