LLMs generate theses kind of graph by writing a python program that plot a real dataset.
Assuming it even correctly does that. And why is that even necessary? People were able to make graphs before AI. It isn’t hard to do, and the piddling amount of data on display hardly needs a custom python program to plot it.
that's also the nature of humans.
Thoughts are more than statistical associations, and anyone telling you that you can make a humanlike consciousness through a glorified autocorrect is a scam artist.