You have that backwards. People are using the colloquial definition of AI.

“Intelligence” is defined by a group of things like pattern recognition, ability to use tools, problem solving, etc. If one of those definitions are met then the thing in question can be said to have intelligence.

A flat worm has intelligence, just very little of it. An object detection model has intelligence (pattern recognition) just not a lot of it. An LLM has more intelligence than a basic object detection model, but still far less than a human.