President Donald Trump announced to the world this week that the United States will begin using the term “super intelligence” instead of AI. This could complicate discussions of the technology, as “superintelligence” has a specific meaning in the AI industry.
The word “artificial,” he said in a speech to the UN General Assembly Tuesday, “makes it sound fake, and it is not fake, it’s actually amazing.”
Download the Straight Arrow app today to get the stories that matter free from manipulation, bias or agenda.™
Point phone camera here
Representatives from the U.S. used the term in their UN speeches all week, but other countries have chosen not to follow suit.
So, what do these terms mean? And what clarity do we lose when we change labels in the industry formerly known as AI?
READ MORE: A nuclear moment? Fractures show at UN Security Council meeting over AI safety
‘Artificial,’ ‘super’… what?
“Artificial intelligence” is a broad research field in computer science and mathematics that focuses on training computers to do tasks.
It was coined in the 1950s by Dartmouth computer scientist John McCarthy, who wanted a catchy term for the field to attract funding.
The “artificial” refers to the fact that the “intelligence” comes from a series of computer calculations, rather than a human or living creature.
What we now commonly refer to as “AI” has expanded to include commercial products like Gemini and ChatGPT, as well as other generative and predictive models and image generators.
The term “superintelligence” is a little more complicated, which means that this federal rebrand might confuse people.
In 2014, then-Oxford philosopher Nick Bostrom published “Superintelligence,” which discussed theoretical outcomes of AI one day becoming smarter than humans in all fields. The book “warned humanity that the creation of superintelligent AI is likely to be our final act — for good or ill,” a forerunner and foundation for recent claims that AI could lead to the apocalypse.
“The whole fear stems fundamentally from this idea that you will reach at some point an “intelligence explosion” where AI will improve itself dramatically and can continue to improve,” Zack Korman, cybersecurity expert and CEO of the AI company Embroidery, told Straight Arrow.
“So, you’ll go to bed one night and it will be really smart and you’ll wake up the next morning and it will be like a God.”
Trump’s federal rebrand of the whole AI industry as “superintelligence” could be confusing, Korman said, because, while the technology is sophisticated, it is not all-powerful. On the other hand, it could be good, “because it kind of makes the whole distinction meaningless.”
And “maybe that distinction should be meaningless, because maybe superintelligence like in the scary God-like sense is not real,” he said.
READ MORE: Are we all going to die? Inside the push to control AI regulation
God-like ‘intelligence’
Bostrom’s theories are contested. His 2014 book built on his earlier philosophizing about how evolving technologies could lead to a future in which humans either don’t exist, or merge with machines.
“In the case of radically transforming technologies, a better understanding of the transition dynamics from a human to a ‘posthuman’ society is needed,” Bostrom wrote in a 2002 paper.
That research has informed the “transhumanist” movement, which suggests that humanity could be generally improved by merging with a god-like superintelligence and with technology in general. One famous transhumanist is millionaire Brian Johnson, whose public attempts to age backward and achieve immortality have included blood transfusions from his son, and who also recently announced his diagnosis of a rare autoimmune gastritis.
Bostrom was fired from Oxford in 2024 after his research institute was closed. This followed the publication of a 2003 email in which he “used the N-word and argued that white people were more intelligent than Black people,” according to The Guardian.
Hang on, what about ‘AGI’?
People sometimes use “superintelligence” and “artificial general intelligence” interchangeably.
But “AGI,” as it is known, is also a slightly different thing. It refers to a type of AI that is intelligent to the level of a human and can do all human-level tasks. The term was popularized in the early 2000s by the researchers Ben Goetzel and Shane Legg.
Legg is one of the founders of Google’s AI DeepMind Institute, and leads its safety team. He recently announced he believes the company is close to achieving AGI, describing it as “a system that exhibits all the cognitive capacities of a human brain.” He admitted that AI “can still sometimes fail at basic tasks and lacks the consistency and creativity to meet the bar of full AGI,” he posted, but “we expect those gaps to be closed soon.”
Goetzel is a transhumanist, as well as the CEO of an AI research company called SingularityNET. He has come under scrutiny for his relationship with Jeffrey Epstein, including telling Epstein that AGI would “supercede” the societal problems caused by “human brain architecture” that had resulted in Epstein’s 2008 jail sentence, and sending him a research funding proposal to build “a robot toddler with commonsense understanding of its environment” that could theoretically use AGI to learn even more.
AI systems are becoming more technically capable, but researchers continue to debate whether those capabilities can be considered “human-level,” earning the label “AGI.”
So… Is AI super-intelligent?
AI is extremely capable, Korman said. But he emphasised that recent announcements from AI companies that their products had “escaped” or “gone rogue” should not be considered evidence of“superintelligence.”
“Understand that if you work in cybersecurity, every day for the last 20 years has been ‘this thing escaped and hacked,’” he said. “The AI story here is not that dramatically different.”
Being an expert cybersecurity defender takes the same skills as being an expert attacker: knowing where all the vulnerabilities are. AI models are expected to be good at breaking out of sandboxes and safety environments because they are trained to write secure code, Korman explained.
READ MORE: Publicity stunt or Mea Culpa? OpenAI’s latest press release splits the crowd
“It’s like a known area of cybersecurity that we were focused on,” he said. “You can’t let it run or it will go hack companies.”
For example, in the case of the “Hugging Face incident,” he said, OpenAI was at fault for not running its test correctly. “The fact that it broke out of OpenAI sandbox, t was a bad sandbox, and they weren’t running monitoring at all. Like they did not run monitoring.”
But, for people who believe a god-like super-intelligence AI will one day take over, “then of course this is like the prophecy coming true,” he said. “They’re kind of allowing their own mistake to become proof of their prophecy.”
Korman finds it annoying to rename the technology at the national policy level. But, he said, “it could be good-to-neutral, because it cuts off the argument that super intelligence implies God.”
Round out your reading
- He built his own Flock-style camera to track police. The police didn’t like it.
- Trump’s apparent joke about letting Iran bomb LA and San Diego catches bipartisan condemnation.
- Grandparents are fighting for access to their grandchildren. Here’s what the law says.
- When the ‘victim’ is made with AI, child pornography cases get complicated.
- Fibermaxxing is trending. Here’s what the science says.
