The basics
Artificial general intelligence (AGI) is a hypothetical AI that could learn and perform any intellectual task a human can, rather than only specific tasks.
Today's AI is narrow. A system that beats champions at chess cannot fold laundry, and a chatbot cannot learn to drive by practicing. AGI would be different: one system flexible enough to pick up almost any mental task, transfer skills between them, and keep improving, much like a person can.
As of 2026, AGI does not exist. Modern language models are impressively broad. Some researchers argue they are early steps toward it. Others point out that they still fail at tasks children find easy, reason unreliably, and cannot learn from experience the way people do. There is no agreed test for when AGI has arrived.
AGI is often confused with superintelligence, which would exceed human ability, and with consciousness, which is about feeling. AGI is only about capability. It is a research goal and a topic of serious safety debate, not a product you can use.
A true AGI could pass your chemistry test in the morning, repair your bike in the afternoon, and plan a family trip in the evening, without special training for each.