Answer engine summary
Artificial general intelligence is a hypothetical system able to transfer competence across unrelated tasks, in contrast to the narrow systems that exist today.
The definition
Artificial general intelligence describes a system that can apply what it has learned in one domain to a genuinely different one, without being rebuilt or retrained from scratch for each.
The emphasis is on transfer. Breadth achieved by assembling many narrow capabilities is not the same thing, though it can look similar from outside.
What makes it hard to assess
There is no agreed test. Proposals range from economic benchmarks to open-ended problem solving, and each tends to reflect a particular view of what intelligence is for.
This is why announcements about progress towards AGI are difficult to evaluate: the goalposts are not merely moving, they were never firmly planted.
Why the term is contested
Some researchers consider AGI the central goal of the field. Others regard it as poorly defined and a distraction from measurable capability and present-day harm. Both positions are held by serious people.
Why fiction still uses it
Because the uncertainty is the point. A concept that cannot be pinned down is difficult for policy and excellent for novels, which can hold an open question for three hundred pages without needing to settle it.
“The record changed, and the room accepted it.”
Questions
Do current AI systems qualify as AGI?
No. They remain narrow, and there is no agreed evidence of the transferable generality the term describes.
When will AGI arrive?
There is no reliable answer. Expert estimates vary enormously and this page does not endorse any of them.
Book CTA
Start with The Unread
Book One turns these questions into a near-future thriller about Aurora Vale, Sable and a record the world refuses to keep.