“It’s really good, like materially better.”
CEO Eyes Only
OpenAI has apparently been demonstrating GPT-5, the next generation of its notorious large language model (LLM), to prospective buyers — and they’re very impressed with the merchandise.
“It’s really good, like materially better,” one CEO told Business Insider of the LLM. That same CEO added that in the demo he previewed, OpenAI tailored use cases and data modeling unique to his firm — and teased previously unseen capabilities as well.
According to BI, OpenAI is looking at a summer launch — though its sources say it’s still being trained and in need of “red-teaming,” the tech industry term for hiring hackers to try to exploit one’s wares.
That last part is important because, as the same website reported just shy of a year ago, GPT-4 had a major race problem prior to being “red-teamed” by OpenAI’s expert exploiters.
Despite OpenAI’s seemingly laissez-faire attitude about the LLM’s unscheduled release date, however, complaints about GPT-4’s apparent degradation have been stacking up in recent months as the model turns a year old. Folks within the company, BI‘s sources say, are hopeful that the release of GPT-5 and introduction of its impressive capabilities will quell those grumblings.
Indeed, even OpenAI CEO Sam Altman has taken to trashing his company’s latest publicly available model in a lengthy and wide-ranging interview with MIT researcher-cum-podcaster Lex Fridman.
“I think it kinda sucks,” the firm co-founder told Fridman of GPT-4.
TBA
In that same interview, the podcaster asked Altman if he could provide a ballpark release date, to which the CEO gave some rather cryptic answers.
“We will release an amazing new model this year,” Altman said. “I don’t know what we’ll call it.”
When Fridman pushed, Altman expounded — sort of.
“We’ll release in the coming months many different things,” he continued. “I think before we talk about a GPT-5-like model called that, or not called that, or a little bit worse or a little bit better than what you’d expect from a GPT-5, I think we have a lot of other important things to release first.”
During the same interview, Altman notably also refused to answer any questions about OpenAI’s secretive Q* project, which was said to be linked to the attempted coup last November which saw him sacked and subsequently rehired — and which makes the vagueness surrounding GPT-5, or whatever it’s going to be called, all the stranger.
More on open questions: OpenAI Says Sora Video Generator May Allow Nudity
Share This Article