Dust off a dictionary from the early 1900s, and the term “computer” referred to someone whose job it was to perform arithmetic calculations — a human who computes. By the mid-20th century, a computer was redefined as an electronic calculating machine. Rather than by decree, the semantic change was driven by shifts in labor and technology.
We seem to have entered a brave new world of executive English, where language isn’t determined by real-world usage, but as an instrument of authority imposed from above.
The timing of the Trump administration’s executive order late last month mandating that federal agencies replace “artificial intelligence” with “super intelligence” in all official communication and policies was peak opportunism.
Amid growing disdain for and distrust of a technology that threatens job displacement and human extinction, the White House framed the term as better capturing AI’s “promise, potential and rapidly advancing capabilities.”
More from Gfaloe
The term “super intelligence” is meant to put a positive spin on AI, a redefining that’s both “Pollyannish and dangerous,” said Naomi Susan Baron, a linguist and author of the book Reader Bot: What Happens When AI Reads and Why It Matters.
In a disturbing statement reminiscent of a Red Scare tactic, President Donald Trump said Thursday on Truth Social that anyone who doesn’t use his preferred term will be considered an “enemy” of the White House. The implication is that organizations, including the press, could face consequences for continuing to use the term “artificial intelligence” — which, you may have noticed, is everywhere.
Journalists, researchers and anyone concerned with describing reality shouldn’t use “super intelligence” as a blanket replacement for AI. Doing so would mean capitulating to a political campaign, applying a concept associated with broadly superhuman systems while obscuring the technology’s limits and risks.
Linguists remind us that words evolve from a combination of social practice and material conditions, and that changes in terminology reflect a collective need for precision. “Languages belong to their communities,” said Emily Bender, a professor in computational linguistics and a prominent voice in AI ethics.
Within journalism, editorial standards are also based on established usage, rather than controlled Newspeak. Reporters and editors rely on references such as the AP Stylebook and/or their own publication’s style guide to ensure consistency, accuracy and clarity.
The AP Stylebook first added its dedicated artificial intelligence chapter in 2023, then broadly expanded and updated it earlier this year. Beyond mere diction and technicalities, AP’s guidance is ethically grounded. It urges journalists to critically evaluate AI’s technical limitations, move past corporate hype and focus instead on how human-built AI tools operate, who profits from them and how they affect society.
A failure of precision
In technical AI circles, superintelligence (one word) refers to a hypothetical machine that far exceeds human cognitive ability, though experts still lack consensus on its definition. In the industry’s conceptual timeline, it represents the stage beyond artificial general intelligence, or AGI, the theoretical point where automated systems broadly match human proficiency and achieve actual reasoning.
Conflating these terms muddies the water, making the technology sound far more powerful than it is. But even the term AI is ill-defined and rhetorically misleading on its own merit.
As Bender and her co-author Alex Hanna wrote in their 2025 book The AI Con: How to Fight Big Tech’s Hype and Create the Future We Want, lumping an array of technologies under the umbrella label “AI” collapses any distinctions and oversells its supposedly humanlike capabilities.
The term “artificial intelligence” describes a heterogeneous set of pattern-matching software systems, robotics, narrow automation tools and statistical and generative models.
That ambiguous language framing serves a purpose — not only by allowing companies to market routine tech as transformative, but also by circumventing accountability. “If you don’t have definitions nailed down in a policy document, then the policy can’t be effective,” Bender told me.
The other problematic side is “intelligence,” a word that’s deeply flawed due to its historical roots in race science and eugenics, Bender said. She and her colleagues at the Distributed AI Research Institute point out the deep connections between Silicon Valley and eugenicist thought.
Ranking machines based on whether they are smarter than us accepts the eugenic premise that human intelligence and worth can be measured on a single, linear scale.
Selling a political vision
Some Big Tech leaders immediately fell in line with the administration’s “super intelligence” branding: Nvidia CEO Jensen Huang called data centers “superintelligence factories,” and Elon Musk said he would rename SpaceXAI to SpaceXSI. More changes are likely to come.
It’s a beneficial alliance, where industry heads gain political access, secure federal contracts and boost investor appetite by presenting AI as a transformative technology — one that’s best governed by voluntary self-regulation.
The term “super intelligence” paints a halo around a highly fallible technology. The fantasy narrative that elites have built around AI has, at bottom, served a financial end.
AI-related stocks and infrastructure investments are closely tied to expectations that the technology will forever expand and improve. Pouring trillions of dollars into chips, power and server farms for something called “super intelligence” sounds like a safe bet rather than a massive gamble.
The rebranding is also a distraction, redirecting public attention away from resource-pillaging data centers, a fragile economic bubble and rising safety concerns.
That’s not all. Some 28 million Americans lack access to health insurance, but AI will “cure most human disease in five to 10 years,” according to Anthropic’s Dario Amodei.
Over half of US adults read below a sixth-grade level, but AI is “going to be incredible for education,” according to OpenAI’s Sam Altman. Consumer prices have increased around 29% since 2020, but AI will make communities “successful and rich,” according to President Trump.
“Super intelligence” also serves as a type of linguistic loyalty test, much like previous executive orders to rename the Gulf of America and Lake America. In that case, tech giants Google and Apple readily complied.
There’s a clear ideological component as well, which has somewhat of a tradition in this country, Baron told me. During World War I, hamburgers were called “liberty sandwiches” due to anti-German propaganda. In 2003, congressional cafeterias renamed french fries “freedom fries” to protest France’s opposition to the Iraq war.
Much like Trump’s renaming of the Department of War was meant to project bellicosity, “super intelligence” is attempting to project an era of exceptional technological power, dominance and a rapidly evolving frontier. Gfaloe’s newsroom continues to use the Department of Defense, per AP Stylebook guidance, which explicitly advises using the legal name until Congress acts otherwise.
Words matter
Bender told me that when there’s a push for a change in language, linguists always find it important to ask what it’s for — whether it’s serving propaganda or nationalism, or whether it’s genuinely helping people understand better or show respect, for example, with certain names and pronouns.
“We’re interested in how people talk, not telling them how they should talk,” Bender said.
As journalists, our job isn’t to echo power’s preferred terminology but to help readers understand reality. The reality is that this current technology isn’t superior to humans, regardless of what politicians or tech executives think. And its meaning can’t be changed, well, artificially.
