Sam Altman Shuts Down Q* Questions


“It’s very mysterious, Sam.”

Loose Lips

OpenAI CEO Sam Altman is remaining tight-lipped about the company’s secretive Q* project — even after admitting that his company is something of a leaky ship.

Even among those who followed along with OpenAI’s November massacre that saw Altman temporarily ousted, it’s easy to overlook the Q* (pronounced “queue-star”) of it all, particularly because nobody outside the company really knows what the heck it is.

The speculation-laden project was linked to the chaos at the firm in the aftermath of that failed coup, and at the end of 2023, OpenAI refused to answer any questions about it — a posture Altman continued in a new interview with tech podcaster Lex Fridman.

When speaking with the high-profile MIT researcher, Altman was initially jokey in his attempts to dismiss any suggestion that the firm he co-founded alongside Elon Musk is hiding anything of import when it comes to the mysteriously-named project.

“OpenAI is not a good company at keeping secrets,” Altman told Fridman. “It would be nice if we were able to have something like that.”

But when the podcaster-cum-MIT researcher pressed him on what Q* actually is, his jovial tone shifted.

“We are not ready to talk about that,” Altman responded curtly.

“See, but an answer like that means there’s something to talk about,” Fridman retorted. “It’s very mysterious, Sam.”

Within Reason

In an attempt at deflection, the CEO made vague statements about OpenAI’s “work on all kinds of research” before trailing off when discussing “better reasoning” in AI systems.

“We haven’t cracked the code yet,” he continued. “We’re very interested.”

Those admissions appear to be along similar lines to others we’ve heard about Q* since learning of its existence during the “turkey-shoot clusterfuck,” as folks at Microsoft called last fall’s debacle. But on closer inspection, the terse refusal to admit to anything other than its existence is actually pretty telling.

As The Information reported during the kerfuffle, Q* is said to “be able to solve math problems that it hadn’t seen before” and provide advances in AI functioning the likes of which the world has never seen. In subsequent reporting, Reuters noted that insiders claim the model is capable of “performing math on the level of grade-school students.”

When lining up those sorts of clues, which Ars Technica powerfully pieced together last December, it appears that OpenAI is indeed sitting on a massive breakthrough in computational reasoning — but just how huge a leap it will be remains to be seen. For now, Altman being coy does nothing but build the hype.

More on OpenAI: OpenAI Says Sora Video Generator May Allow Nudity


Leave a Reply

Your email address will not be published. Required fields are marked *