Artificial general intelligence refers to a proposed AI system with broad competence across many domains rather than strong performance on a narrow task set. Definitions commonly emphasize learning new tasks, transferring knowledge, adapting to unfamiliar situations, and reasoning across different problems.
There is no universally agreed test or threshold for declaring that such a system exists. Claims should identify the definition and evidence being used instead of treating one person's interpretation or a collection of demonstrations as a settled standard.
ELI5
Artificial general intelligence is a proposed kind of AI that could learn and perform a broad range of cognitive work instead of being strong only in narrow tasks. It would transfer knowledge across domains and adapt to unfamiliar problems at a human-comparable or greater level.
For example, doing well on one coding test would not by itself demonstrate general intelligence. A credible claim would need clear definitions and evidence across many different tasks, environments, learning situations, and real-world constraints.
Is there an agreed standard for artificial general intelligence?
No. Researchers and organizations use different definitions, thresholds, and views about the evidence that would be sufficient.
Is broad task performance enough to establish general intelligence?
Not automatically. Evidence should cover adaptation, transfer, reliability, unfamiliar tasks, and the limits of the evaluation.
