Musk is no stranger to an existential debate, and with xAI, he intends to solve fundamental scientific problems surrounding gravity, dark matter, the Fermi Paradox, and more.
On the 15th of July, 2023, AI critic Elon Musk presented his latest brainchild, ‘xAI,’ during an interactive Twitter Spaces event.
His mission? “To build a good AGI with the overarching purpose of just trying to understand the universe,” Musk said. Later, he added, “I guess you could reform the xAI mission statement as: ‘What the hell is really going on?’ That’s our goal.” Artificial general intelligence (AGI) is AI exceeding human intelligence.
Musk explained that xAI aims to untangle universal enigmas and existential questions – from dark matter and dark energy to the Fermi Paradox and the existence of aliens. He intends for xAI to probe “fundamental questions” concerning “reality” and “the universe.”
xAI Discussion https://t.co/UHZatW4vFS
— Elon Musk (@elonmusk) July 14, 2023
That’s a tall order – it would likely require AI to escape the confines of its current architectures and the nature of generating answers from training data.
Currently, popular AIs, such as Bard and ChatGPT, can only return summaries and analyses of human-written facts, debates, and opinions surrounding such issues. In other words, while current generative AI can synthesize information from different sources, it can’t generate novel or pioneering theories, and its outputs are fundamentally derivative.
Musk wants to change that. His team hinted that xAI could be more mathematically-driven than current leading AI products, which would be necessary to support scientists in solving highly complex problems.
xAI’s cosmic ambitions
Dark matter, dark energy, gravity, and aliens – a handful of the scientifically contentious topics Musk intends xAI to help solve.
Musk spoke about The Fermi Paradox, the contradiction between the high probability of extraterrestrial life, suggested by the vast number of stars and planets in the universe, and the lack of evidence or contact with such civilizations.
“If anyone would have seen evidence of aliens, it’s probably me. And yet, I have not seen even one tiny shred of evidence for aliens,” he said, explaining we “may actually be the only thing, at least in this part of the galaxy.”
Musk has previously responded to questions surrounding alien life and the Fermi Paradox by arguing we’re living in a simulation, also called simulation hypothesis.
In 2016, at Recode’s Code Conference, Musk famously said, “There’s a billion to one chance we’re living in base reality.”
The essence of the simulation hypothesis is that we’re currently living in a computer simulation, perhaps built by sophisticated ancient aliens or our predecessors before our planet was destroyed. Musk has argued that current advancements in AI technology support simulation theory.
Someone joked on Twitter that if we are living in a simulation, then xAI is Musk’s plan to become Neo from the Matrix.
Igor Babuschkin, who worked on the Large Hadron Collider at CERN, shared Musk’s ambition to answer existential scientific questions at xAI.
The company wants to “really make a dent in our understanding of the universe,” Babuschkin said.
Jimmy Ba, an AI researcher from the University of Toronto professor, echoed these sentiments, stating, “How can we build a general purpose problem-solving machine to help all of us, humanity, overcome the most challenging and ambitious problems out there, and how can we use those tools to augment ourselves and empower?”
Musk spoke of his ambition to build AGI. “So there are all these, like, fundamental questions. I don’t think you can call anything AGI until it’s solved at least one fundamental question,” he remarked.
“If I could press pause on AI or really advanced AI digital superintelligence I would. It doesn’t seem like that is realistic so xAI is essentially going to build an AI … in a good way, sort of hopefully.” The ‘hopefully’ there is perhaps indicative of Musk’s understanding of the perils of pursuing such goals with AI.
Musk envisions the creation of an AGI that will embody “maximal curiosity” and a “truth-seeking” approach devoid of political correctness. He suggests that xAI’s responses might stir controversy, even if they hold factual value.
However, xAI’s ambitions are not purely academic. The xAI team, consisting of experts previously associated with giants such as Microsoft and Google, highlighted the critical issue of “monopolization” in AI. They expressed concern that the current AI incentives are “not aligned with the rest of humanity,” which could be a significant risk.
Back in April, Musk accused Microsoft and OpenAI of “training the AI to lie” and argued OpenAI has now become a “closed source,” “for-profit” organization “closely allied with Microsoft.” This is true – OpenAI engages in virtually no open-source collaboration like it once did.
How xAI will achieve its goals
Musk hinted that xAI would collaborate closely with both Twitter and Tesla.
“We are going to work with Tesla on the silicon front and maybe on the AI software front as well,” he said, referring to Tesla’s proprietary chips. However, he highlighted that xAI’s collaboration with Tesla is complex due to investment interests and its public status.
In terms of the technical nitty-gritty of building an AI that can answer fundamental scientific questions, Musk didn’t offer much insight.
He conceded that his company wouldn’t need as much computing power as other AI giants like OpenAI. Babuschkin indicated that the team would remain relatively small.
Musk also said they’d likely use Twitter data for AI training, noting that many other companies are illegally scraping their data already.
We don’t yet know how this model will be built, on what architectures, or whether it’ll be publicly accessible or purely experimental for scientific use. However, Musk said xAI was “definitely in competition” with OpenAI – so surely they’d need to release publicly accessible products?
The team noted that further details on xAI – perhaps including some technical strategy regarding modeling, architectures, training, etc – will be available within 2 weeks.
Musk on AI: a timeline
xAI is shadowed against a decade-long backdrop of Musk’s criticisms of AI.
Here’s a brief synopsis of Musk’s views on AI.
- August 2014: Musk alerts the world, “We need to be super careful with AI. Potentially more dangerous than nukes.”
- October 2014: He doubles down on the warning, stating, “I think we should be very careful about artificial intelligence… our biggest existential threat.”
- October 2014: In a comparison, Musk claims, “With artificial intelligence, we are summoning the demon.”
- June 2016: He speculates about a future where AI outpaces human intelligence, likening humans to pets.
- July 2017: Musk calls AI a “civilization-level risk” and stresses the need for safety research.
- July 2017: He admits to having exposure to advanced AI and advocates for concern over it.
- August 2017: Musk urges people to be concerned about AI safety, comparing its risk to North Korea.
- November 2017: He grimly estimates a mere 5-10% chance of making AI safe.
- March 2018: Musk calls for regulatory oversight of AI, which he believes is more dangerous than nuclear weapons.
- April 2018: He warns that AI could lead to the creation of an “immortal dictator.”
- November 2018: Musk humorously muses about AI forcing him to submit to it.
- September 2019: He predicts using advanced AI to manipulate social media.
- February 2020: At Tesla, he speaks about the integral role of AI in self-driving technology.
- July 2020: Musk anticipates that AI will vastly outperform human intelligence in less than five years.
- April 2021: He asserts that real-world AI is necessary for unsupervised, generalized full self-driving.
- December 2022: Musk warns about the potential dangers of training AI to align with ‘woke’ culture.
- 2023 to date: Musk has debated AI with numerous prominent figures on Twitter, typically arguing that the technology will destroy humanity.
- June 2023: Musk founded an AI company intending to solve existential and fundamental scientific questions.
Musk has a reputation for speaking in platitudes lacking in substance, and his views on fundamental scientific questions have snaked from one theory to another. His public notoriety and cult-like status peaked in 2018 when he swigged whisky while puffing a marijuana joint on a podcast with Joe Rogan.
Many argue that xAI is an intellectual project designed to enhance Musk’s visionary ambitions rather than delivering genuine value.
The day after his discussion on xAI, Musk’s Falcon 9 rocket launched 54 Starlink satellites into orbit – there’s no criticizing his immense appetite for futuristic appetite for innovation.
Regardless of anyone’s view on Musk himself, xAI has been chucked into the AI industry as a wild card, and Musk has assembled a prolific team to back him.
xAI’s true intent and methodology will be revealed over the course of this year.
While Musk’s willingness to jump on the AI bandwagon in pursuit of what he deems a morally just cause has sparked controversy, there’s no doubt that xAI is a gripping entrant into the market.
The prospect of a mathematically-driven AI model designed to solve complex scientific questions is tantalizing – and Musk has assembled an eclectic team to back his visions.
However, for now, we simply need more detail to understand how xAI will answer, “What the hell is really going on?”