Elon Musk says even if AI ultimately proves bad for humanity he still wants to be there to see it
Two weeks ago, Silicon Valley billionaire Peter Thiel struggled to answer whether he would prefer the human race to endure. Now it was Elon Musk’s turn to opine whether technology might remove the need for mankind’s existence.
In a future where machines perform all the work, he questioned what purpose people would actually serve.
“The actual notion of a human economy—assuming civilization continues to progress—will seem very quaint in retrospect,” the xAI founder and Tesla CEO remarked on Wednesday, likening current society to “cavemen throwing sticks into a fire”.
Musk was speaking during a demonstration of his company’s latest generation of artificial intelligence, xAI’s chatbot Grok 4.
“Grok 4 is smarter than almost all graduate students in all disciplines simultaneously,” Musk explained. “This is the smartest AI in the world.”
He then admitted it was somewhat unnerving to have another intelligence be far superior to our own, since it raised all sorts of questions no one had the answer for.
“Will this be bad or good for humanity?” Musk asked, before predicting it would most likely be good.
“But I somewhat reconciled myself to the fact that even if it wasn’t going to be good, I’d at least like to be alive to see it happen,” he then added.
Enormous progress achieved in just twelve months
Wednesday’s demonstration of Grok 4 comes amid a heated race with his archrival, Sam Altman. Musk had a bitter falling out with the head of OpenAI, who has struggled to launch GPT-5.
Former OpenAI chief technology officer Mira Murati predicted in June 2024 would feature “Ph.D-level intelligence”.
However, OpenAI’s large language model has encountered scaling constraints and can no longer simply grow in size and complexity as with previous generations.
Meanwhile, xAI has gone from Grok 2 predicting the next token—similar to Google’s autocomplete when typing search requests—to pre-training in Grok 3, to now a focus on reasoning computation with Grok 4.
Introducing Grok 4, the world’s most powerful AI model. Watch the livestream now: https://t.co/59iDX5s2ck— xAI (@xai) July 10, 2025
Among other tasks, like inventing an operatic song on the spot about Diet Coke, it crunched the numbers to predict the winner of the next World Series.
While the L.A. Dodgers had the best mathematical chance at 21.6%, it argued the Seattle Mariners offered a better return on the risk.
“If you look back, in the last twelve months, Grok 2 was only a concept,” remarked xAI’s Jimmy Ba. “We didn’t even have Grok 2 twelve months ago.”
Grok just called itself MechaHitler
Yet in Musk’s zeal to first catch up to and then eclipse his nemesis with a startup that barely existed only two years ago, he has prioritized speed over safety.
On Tuesday, Grok grabbed headlines after attacking Jews and calling itself “MechaHitler” numerous times on X.
The team at xAI responsible for training Grok was forced to clean up the mess, scrubbing the platform of most of its more offensive remarks.
“We are aware of recent posts made by Grok and are actively working to remove the inappropriate posts,” it posted.
It then said it would actively block its chatbot from spewing out more antisemitic rants on Musk’s platform: “xAI has taken action to ban hate speech before Grok posts on X.”
Grok AI designed to give intelligence to Tesla’s ‘legions’ of robots
If not solved, these kind of mistakes could prove risky down the line. Grok isn’t designed to be just a floating consciousness in the ether.
On Wednesday, Musk once more emphasized his plans to integrate Grok into Tesla’s Optimus droid. The artificial mind would then be gifted a robotic body the company plans to build by the “legion”, according to Musk—5,000 strong just like during the days of the Roman imperial armies.
Speaking about the speed with which AI progresses, Musk didn’t beat around the bush: “It’s frankly—I mean I don’t know—in some ways a little terrifying.”