AGI stands for artificial general intelligence. It is the name researchers and AI companies use for a system that can match or exceed human ability across most intellectual tasks, rather than excelling at one.
The concept is the same as general AI. This page focuses on how the term “AGI” is used today: who uses it, how it is defined and how people try to measure it.
Where the term came from
The phrase appeared in the late 1990s and became popular after the 2007 book Artificial General Intelligence, edited by Ben Goertzel and Cassio Pennachin. Researchers wanted a name for the original goal of AI, human-like general intelligence, as the field had shifted toward narrow applications.
How AGI is defined today
There is no single definition. Common versions include:
- Capability-based: can do most cognitive tasks at least as well as a skilled human.
- Economic: OpenAI’s charter describes AGI as highly autonomous systems that outperform humans at most economically valuable work.
- Levels-based: a 2023 Google DeepMind paper proposed grading AGI in levels by performance and generality, rather than treating it as a single finish line.
How would we know AGI has arrived?
This is one of the most debated questions in the field. Older ideas like the Turing test are now seen as too easy to game. Newer benchmarks test reasoning on unfamiliar puzzles, real-world work tasks or long, multi-step projects. Each lab tends to favour its own measures, so claims of “AGI” should be read with care.
Why the term matters
AGI is not only a technical idea. It shapes company missions, investment, regulation and public fear or excitement. When a company says it is “building AGI”, ask which definition it means. The answer changes what the claim is worth.