Donald Trump’s push to call AI “super intelligence” has reopened a question that matters beyond branding: what, exactly, are we claiming when we say a machine is intelligent? New Scientist argues that AI, artificial general intelligence and superintelligence describe different things, and that the goalposts keep moving.
Watch Desk analysis
What happened
The article traces the changing vocabulary around AI, from narrow systems built for specific tasks to artificial general intelligence, or AGI, usually understood as human-level performance across a broad range of tasks. It distinguishes both from superintelligence: philosopher Nick Bostrom’s term for an intellect that greatly exceeds human cognitive performance across virtually all domains.
Trump has called for AI to be rebranded as “super intelligence”; the article says an executive order signed on 29 September used that term and a new “Super Intelligence Force” was launched to coordinate US AI policy. But a change of label does not establish a change in capability. New Scientist notes that current definitions of AGI are woolly, while tests can be overtaken by new versions, shifting benchmarks or disagreement about what intelligence measures.
Read the New Scientist article.
Why it matters
These terms shape how companies describe their ambitions, how people interpret capability claims and how governments frame AI policy. If “superintelligence” is used for today’s AI systems as well as for a hypothetical intellect far beyond human ability, the same word can make very different claims sound alike.
The measurement problem is real too. New Scientist points to the limits of IQ-style tests and the difficulty of deciding whether a benchmark score demonstrates general intelligence. A high score on one test can be meaningful without settling the much larger question of what a system can do across domains.
Our read
The useful takeaway is not that labels never matter. It is that they need definitions attached. “AGI” can mean different things to different organisations, and “superintelligence” makes an even bigger promise. Ask what capability is being claimed, how it was measured and what the test leaves out. A grand name is not a benchmark, however good it looks on a letterhead.
What to watch
- How US officials define “super intelligence” in policy and practice.
- Whether AI companies state clear, testable definitions of AGI.
- How new benchmark versions change the meaning of headline scores.
Discussion spark: Should governments and AI companies be expected to use agreed definitions for terms such as AGI and superintelligence, or would that give a false sense of precision to a moving target?
Sources and evidence
- AI or ‘super intelligence’? Trump isn’t the only one who is confused (8 October 2026, 16:07 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.