AGI

AGI (Artificial General Intelligence) is the hypothetical point at which an AI can handle essentially the full range of intellectual tasks a person can, and carry what it learns on one task over to another, instead of being built for a single job.

The contrast is narrow AI: today’s systems, including GPT -style large language models, are remarkably capable but still tied to the kinds of tasks and data they were trained on. There’s no agreed definition or test for AGI. Proposals range from “matches a skilled human at most economically valuable work” to “can do anything a human can,” and experts genuinely disagree on whether simply making today’s machine learning models bigger will get us there or stall short of it.

A related term, ASI (Artificial Superintelligence), means a system that would outperform the best humans at everything. In practice “AGI” is a moving, contested goalpost, and a marketing word about as often as a technical one, so when someone uses it, it’s worth asking which specific ability they actually mean.