More open ended range of information fed to the ai

I have been trying to generate some characters from recent media but every time I do it just spits out a flower, or food, or just blocks with the name on it. So I think that adobe should widen the range of information the AI can access.