Artificial General Intelligence (AGI) refers to a system that can match or exceed human performance across the full range of cognitive tasks, transferring what it learns from one domain to another the way a person does, instead of being a specialist that is strong at one thing and helpless at the next. It is contrasted with the “narrow” AI that dominates practice today, where a model that plays a game or predicts a protein cannot, on its own, do your taxes or plan a research programme.
The term is deliberately fuzzy, and definitions differ: some frame AGI around economic substitutability (it can do most of what a remote human worker can), others around cognitive breadth (it exhibits all the capabilities a human brain does). Because there is no single agreed benchmark, claims about “how close” AGI is often turn on which definition the speaker has in mind, which is worth keeping in view whenever timelines are debated.