Elon Musk says he’s embracing the term “Super Intelligence” so completely that he’ll even rename one of his companies for apparent consistency. After announcing his preference for the term Super Intelligence on X, a user asked him about “renaming to SpaceXSI.” Musk replied “Yes, we will make that change.”
This follows president Trump’s executive order late last month renaming artificial intelligence “Super Intelligence,” and statements indicating he believes AI business leaders have agreed to use his preferred term.
Early on Sunday morning, Musk posted this on X:
“No more AI
SI
It’s better”
He then self-replied with “It always was,” and posted the cover of the Nick Bostrom book Superintelligence: Paths, Dangers, Strategies.
Maybe he’s right in a way. Perhaps “superintelligence”—one word—was always a catchier marketing term.
“Artificial intelligence” was an observation from about 1956 of attributes—both real and hypothetical—in computer systems that gave them what one group of researchers thought was the potential to simulate every aspect of intelligence. Over the decades, artificial intelligence has become an evocative term in sci-fi, an area of scientific inquiry, and a marketing term used by, for instance, Zojirushi to describe a flexible form of rice cooker automation.
While that lexical evolution was underway, techniques like deep learning, and architectures like transformers and LLMs slowly caused the concept of AI as understood by computer scientists to morph into a specific set of tools. Today, practically speaking, someone knows what’s meant by AI when they, say, go to buy some of it: usually model-based, generative, token-based software.
Superintelligence, however, is a loosely related term. According to the Oxford English Dictionary, it can mean many forms of superior intelligence, but it primarily means God’s plan, or something like a mysterious underlying order to the universe. The first ever use of the word, from an author named R.H. Black reads, “Providence expresses the particular act of providing..[and] is applied to the Divine Superintelligence.” You have to admit Musk is arguably right: the mysterious power of God has way more juice than a term that has come to mean tokens and LLMs.
Bostrom’s book is more or less about the overlap of AI and superintelligence—a hypothetical point at which computer-based “intelligence” outclasses humanity, and how that could have upsides and downsides. He includes helpful thought experiments that have been absorbed by many overzealous readers as prophecies of doom, which, I’ve learned from experience, is something that often happens when authors speculate.
Mark Zuckerberg and Meta have been using the compound word version of the term “superintelligence” in recent years, largely to describe an aspiration. In his essay from last year, “Personal Superintelligence,” Zuckerberg wrote, “The rest of this decade seems likely to be the decisive period for determining the path this technology will take, and whether superintelligence will be a tool for personal empowerment or a force focused on replacing large swaths of society.”
Trump’s own explanation for his use of the two-word phrase “Super Intelligence” has little to do with marking a major milestone in AI research. His executive order does hint at that milestone without much fanfare, dubiously claiming it can do “much more than imitate or automate discrete aspects of human intelligence.” But off-the-cuff, Trump clearly contradicts this, expressing dislike of the negative connotations of the word “artificial,” and saying, “It’s a terrible word. It’s amazing it lived on so long actually.”
As I wrote before, the business leaders who attended Trump’s AI business summit last week are slowly coming to terms with what they signed up for. I’m not alone in noticing that these businessmen are cornered, and probably a little embarrassed. Writing in Wired, Steven Levy described the debacle as a bunch of CEOs and company presidents—leaders from Anthropic, Meta, OpenAI, Nvidia, and Google—humiliating themselves in order to pass a Donald Trump loyalty test. “My bet is that they will do the minimum to avoid presidential ire while trying to run out the clock until the next administration puts this bizarre idea out of its misery,” Levy wrote.
It’s not at all clear how much—if at all—Trump is pressuring these business leaders via back channels to dig deeper “Super Intelligence” holes for themselves. Musk and Zuckerberg are clearly fond of the word “superintelligence,” but if they start throwing around that term with serious people in the room, they’ll need to clarify what they mean—whether they’ve decided current AI models are “superintelligent,” or whether they’re participating in Trump’s branding exercise. If it’s the latter, are they going to use the two-word and compound-word versions interchangeably? If so, are they going to invent their own new term for the hypothetical technology currently called “superintelligence?”
Then again, SpaceXAI parent company SpaceX already renamed SpaceXAI once this year—it used to be called xAI. And Musk famously fails to deliver on things he says will happen.
