// NATURE NEWS — SPAZIO & SCIENZA
The science of superintelligence and why AI is still a useful word
Jenna Ahart is a science journalist based in Washington DC.
Search author on:
PubMed
Google Scholar
By most researchers’ definitions, superintelligence is an aspirational category of systems, not a reality.Credit: Andrzej Wojcicki/SPL
‘Superintelligence’ has arrived — at least by one definition of the term.
Will AI really kill us all? The science behind the hype
In an executive order issued on 29 September, US President Donald Trump declared that humanity is in “a new era of Super Intelligence”. The term artificial intelligence is outdated, the order said: “[T]oday’s frontier systems do much more than imitate or automate discrete aspects of human intelligence.” The order also directed federal agencies to use the phrases “Super Intelligence” and “SI” instead of “artificial intelligence” and “AI”.
Trump’s declaration has had a mixed reception. Billionaire entrepreneur Elon Musk’s post on the social-media platform X said: “No more AI. SI. It’s better,” adding that his company SpaceXAI will change its name to SpaceXSI. OpenAI chief executive Sam Altman, however, continues to use the term AI in his posts.
Trump’s order defines ‘Super Intelligence’ as “the technologies and systems encompassed by the term ‘artificial intelligence’”. But researchers were using the term superintelligence to refer to hypothetical, highly capable technologies long before the executive order. The word was popularized in science philosopher Nick Bostrom’s 2014 book Superintelligence, which defined it as “any intellect that greatly exceeds the cognitive performance of humans in virtually all domains of interest”.
A 2025 petition, in which a host of AI researchers and others called for a slowdown in the development of superintelligence, similarly describes it as a system “that can significantly outperform all humans on essentially all cognitive tasks”.
But the term is still not completely defined, says Melanie Mitchell, an AI researcher at the Santa Fe institute in New Mexico. For instance, the components of human-level intelligence remain undefined. “None of this has rigorous meaning. People can use these terms to mean what they want,” she says.