What Is Superintelligence? Definition, History, and the 2026 White House Terminology Change
Quick answer: Superintelligence is a term popularized by Oxford philosopher Nick Bostrom, who defines it as "any intellect that greatly exceeds the cognitive performance of humans in virtually all domains of interest." In September 2026, President Donald Trump signed Executive Order 14434, directing federal agencies to use "Super Intelligence" and "SI" in place of "artificial intelligence" and "AI" in official, non-statutory communications. The two usages differ: the federal term carries the same legal definition as AI under existing law, while researchers use "superintelligence" to describe a system that does not currently exist.
What Is Superintelligence?
In research contexts, superintelligence describes a hypothetical system whose intellectual performance would exceed that of the most capable human beings in essentially every field of endeavor. Bostrom, the founding director of Oxford University's Future of Humanity Institute, developed this definition in his 2014 book Superintelligence: Paths, Dangers, Strategies, which examined potential paths to such systems and argued they would be difficult to control.
No system built to date meets this definition. Current frontier models, including large language models and agentic systems, have documented limitations: they fail at tasks humans find straightforward, they require human verification of outputs, and their reasoning remains uneven across domains.
Superintelligence vs. AGI vs. AI
The three terms describe different capability levels:
- Artificial intelligence (AI): Systems that perform tasks associated with human intelligence.
- Artificial general intelligence (AGI): A hypothetical system matching human-level performance across essentially all cognitive tasks.
- Superintelligence (ASI): A hypothetical system exceeding the best human performance across virtually all domains.
The 2026 Executive Order
On September 29, 2026, President Donald Trump signed Executive Order 14434, "Inaugurating the Era of Super Intelligence." The order directs federal agencies to replace "Artificial Intelligence" and "AI" with "Super Intelligence" and "SI" in official correspondence, public communications, websites, reports, policy documents, and other non-statutory documents.
Key provisions, per the order include:
1. Terminology change with unchanged legal definitions. The order states that "Super Intelligence" has the same meaning as "artificial intelligence" as defined in section 9401(3) of title 15, United States Code. It does not alter statutory definitions.
2. A proposed federal definition is forthcoming. The order directs the Assistant to the President for Science and Technology to propose a federal definition within 60 days, along with any recommended conforming amendments to statutory references. (
3. An accompanying industry pledge. On September 30, 2026, the White House announced the "White House Accord on Super Intelligence" (formally the Joint Commitment on Frontier Responsibilities), a voluntary safety commitment signed by executives of Google, Anthropic, Meta, OpenAI, xAI, and Nvidia.
Some observers have raised practical considerations. The International Association of Privacy Professionals noted that the change creates terminology work for governance programs while leaving existing legal definitions in place. Legal analysts writing in Forbes identified open questions around how the rename will interact with existing contracts and regulations.
Divergent Usages
Researchers have noted that the federal usage differs from the research term. Nature reported that the order defines "Super Intelligence" as "the technologies and systems encompassed by the term 'artificial intelligence,'" while researchers use "superintelligence" to refer to a hypothetical, far more capable class of systems that does not yet exist.
For compliance and legal teams, the practical implication is a two-usage environment: federal documents now using "SI" to mean what "AI" meant previously, and technical literature using "superintelligence" for a hypothetical capability level. Distinguishing between the two will matter as agencies issue guidance under the new terminology.
Frequently Asked Questions
What is superintelligence?
A hypothetical intellect that would surpass the best human performance across virtually all cognitive domains at once. Researchers, including Oxford philosopher Nick Bostrom, treat it as a future possibility, not a current technology. (Wikipedia)
Who coined the word?
The modern technical definition comes from Nick Bostrom's 1997 essay "How Long Before Superintelligence?" and his 2014 book Superintelligence: Paths, Dangers, Strategies.
Does superintelligence exist yet?
No system meets the research definition. Stanford HAI's AI Index and other academic assessments treat superintelligence as a hypothetical category. (Stanford HAI)
What did the 2026 executive order change?
Executive Order 14434, signed September 29, 2026 by President Trump, directs federal agencies to use "Super Intelligence" (SI) instead of "artificial intelligence" (AI) in official, non-statutory communications. The legal definition remains unchanged: SI has the same meaning as AI under existing law (15 U.S.C. § 9401(3)). A proposed federal definition is due within 60 days.
Is the federal "Super Intelligence" the same as superintelligence in research?
No. The federal term covers technologies previously described as AI. The research term refers to a hypothetical system vastly exceeding human cognition. The shared wording has created a distinction legal and compliance teams need to track.
How is superintelligence different from AGI?
AGI refers to a system matching human-level performance across all tasks. Superintelligence refers to a system exceeding the best human performance across all domains.
What are the main concerns associated with superintelligence?
Researchers such as Bostrom have described a control problem: a system more capable than its designers might pursue imperfectly specified goals in unintended ways. Other researchers consider these scenarios speculative and prioritize present-day issues such as bias, misinformation, and surveillance. Assessments vary within the field.
