Inveterate overpromiser Elon Musk is teasing that his chatbot Grok could soon “achieve artificial general intelligence,” or AGI — a hypothetical form of AI system that surpasses humans in virtually every way, and remains the white whale of the AI industry.
“My estimate of the probability of Grok 5 achieving AGI is now at 10 percent and rising,” Musk wrote on his website X, formerly Twitter, on Saturday.
Like a good courtier, an employee at Musk’s AI firm xAI was there to boost Musk’s claim within the hour.
“I trust this forecast of probability,” tweeted the employee Aditya Gupta. (This is the same worker who a few months ago was publicly chewed out by his boss for using the word “researcher” instead of “engineer.” At the time, Gupta issued a sheepish correction.)
And thus began a sycophant game of telephone. Musk, in response to someone praising his own prediction, interpreted that as a sign to double down on it — while at the same time hedging his bets, somehow.
“Grok 5 will be AGI or something indistinguishable from AGI,” an emboldened Musk proclaimed.
Musk made a similar bet in 2024, predicting AGI would come “within two years.” He also teased last month that Grok 5 “has a chance” of reaching AGI. But is it a ten percent chance? Or is it something that “will,” as in absolutely, happen, per his latest tweet?
Raising pedantic quibbles like these will make Musk very mad at you, apparently. Because that’s what exactly OpenAI research scientist Gabriel Petersson did.
“10 percent chance Elon declares he reached AGI a fourth time,” Petersson, who works on the company’s video generating AI Sora, joked. “It’s no longer first to AGI, it’s first to 10 AGIs.”
Musk fumed in the replies, reaching back to his strange bugbear about “researchers” versus “engineers” that had previously caused him to rage at Gupta. “You call yourself a ‘researcher,'” he wrote. “Pathetic.”
Musk is in a bitter rivalry with OpenAI, a company he helped cofound in 2015 with its current CEO Sam Altman. Musk departed the firm, which is now being valued at half a trillion dollars, in late 2018 reportedly because of clashing with Altman’s direction. The two have been feuding ever since, with Musk firing off several lawsuits against his former company.
They also have different definitions of AGI: Musk considers it an AI that’s either “smarter than the smartest human” he said in 2024, or “capable of doing anything a human with a computer can do,” he said in a recent tweet (good luck pinning him down on anything.) OpenAI officially defines it as a “highly autonomous system that outperforms humans at most economically valuable work,” but Altman is also on the record saying that AGI is “not a super useful term,” despite continuing to use it.
All that said, it’s a little rich that an OpenAI employee is taking the high ground on AI predictions. Its head honcho Altman has repeatedly exaggerated the capabilities of his own tech and has also hyped AGI as something just around the corner; last year, he proclaimed that AGI is achievable with “current hardware,” for example. In the build up to the launch of OpenAI’s newest GPT-5 model this summer — considered a massive disappointment by many — Altman declared it a “significant step along the path to AGI.” He even called the model “generally intelligent.” It’s the same slimy salesmanship as Musk, only less stupid-sounding.
More on AI: OpenAI Researcher Forced to Delete “Embarrassing” Tweet Claiming Huge Breakthrough