Why is everyone's first example of things you can do with LLMs "revenge porn"? They're text generation algorithms not even image generators. They need external capabilities to create images.
Do you also object to people saying that web browsers "display" a website even though that needs them to be plugged into a monitor?
If you chat to an LLM and you get a picture back, which some support, the image generator and the language model might as well be the same thing to all users, even if there's an important technical difference for developers.
It's a distinction that does not matter, as the question still has to be answered for the other modality. Do guns kill people, or do bad guys use guns to kill people? Does a fall kill you, or is it the sudden deceleration at the end? Lab leak or wet market? There's a technical difference, some people care, but the actionable is identical and doesn't matter unless it's your job to implement a specific part of the solution.
