Generative AI describes output. These are systems trained to produce text, images, audio or code that resembles their training data's distribution. AGI describes scope and autonomy: a system that can set about any intellectual task, including ones it has not seen, and carry it through.
Different questions
Generative AI asks: can the model produce something plausible? AGI asks: can the system decide what needs doing, do it, check whether it worked, and learn from the result? The first is a rendering problem. The second is a control problem.
Where they overlap
Most modern assistants are generative models wrapped in tooling that adds memory, retrieval and execution. That scaffolding moves them toward agency without making them general. It is the difference between a talented improviser and someone who can run a project.
The practical distinction
Generative AI is already here and commercially deployed. AGI remains a target. Confusing the two produces both inflated expectations and premature dismissal.