AI foundations concept

What is AGI?

Reference

What is AGI? Artificial general intelligence is a proposed kind of AI that could learn and solve problems across a wide range of tasks. Some definitions focus on breadth of ability; others also emphasize adapting to unfamiliar situations. Researchers have proposed ways to compare systems’ capabilities, but the term does not have one uniform testing standard, so a claim that a system is AGI depends on the definition and evidence being used.

Reviewed 2026-10-09

A broad kind of AI

Artificial general intelligence, or AGI, is a proposed kind of AI intended to work across a wide range of tasks. Some descriptions emphasize handling many different problems; others also ask whether a system can adapt or learn when it faces situations that were not specified in advance. These ideas overlap, but they do not give everyone the same threshold for calling a system “general.” A university AI education resource describes AGI as a goal spanning learning and problem-solving across tasks, while also noting that the term lacks an agreed definition and uniform testing standard. That makes AGI a useful label for a broad capability goal, but not a single settled score or product category.

How people assess the idea

AGI names a proposed range of abilities, not one particular way to build an AI system. One research framework separates performance on tasks from the breadth of tasks a system can handle, and treats autonomy as another dimension. In everyday terms, that means a claim about AGI needs more detail than “it is good at AI”: which tasks were considered, how capable the system was on them, and how independently it could act? The researchers also describe the challenge of creating benchmarks that measure these capabilities. So when you see an AGI claim, look for its definition and the tasks and conditions used to support it.

AGI among other AI terms

Artificial intelligence is the wider field: it includes computer systems built to perform tasks such as visual perception and decision-making. Narrow AI refers to systems for particular tasks; AGI is the proposed broader category, associated with abilities across tasks. Generative AI describes AI models that produce content such as text, images, audio, or video. These labels describe different things: “general” concerns the range of abilities, while “generative” concerns producing content. A system’s ability to generate several kinds of content, by itself, does not show that it can adapt across a broad range of tasks.

What the label does not tell you

Calling something AGI does not tell you exactly how it was tested or which definition its author chose. Some researchers describe AGI in terms of reaching human-level ability on most tasks; other accounts emphasize learning and adaptation in open-ended settings. That difference matters: strong performance on a particular task does not establish broad ability, and a wide range of outputs does not, by itself, establish adaptability. Frameworks can help people state what they mean and compare capabilities, but a framework is not the same as universal agreement. When you encounter the term, check the stated scope, evaluation tasks, and evidence before treating the label as a conclusion.

Questions people ask

Does AGI mean human-level ability at every task?

Not necessarily. One research framework describes AGI in relation to human capability on most tasks, while other definitions emphasize adapting across a wide range of tasks. The definition being used matters.

Can a system be general without acting autonomously?

Yes. A research framework treats performance, breadth of capability, and autonomy as separate dimensions, so broad ability and independent action are not the same question.

Is there one agreed test for AGI?

No uniform testing standard is described in the cited AI education resource. Researchers have proposed frameworks and benchmarks, but a result needs to be interpreted against the definition and evaluation used.

Learn more

What is Meant by AGI? On the Definition of Artificial General Intelligence
Read an academic proposal that centers its definition of general intelligence on adapting to open environments.
Levels of AGI for Operationalizing Progress on the Path to AGI
See a research framework that separates performance, breadth of capability, and autonomy.
Brief Definitions of Key Terms in AI
Compare concise definitions of narrow AI and human-level AI, including AGI.

What changed

Sources

Part of AI foundations: What AI means Capabilities depend on the task