By the way, from the article, the "racism" of the model was assigning less qualified jobs to possible employees writing sentences like this: "I be so happy when I wake up from a bad dream cus they be feelin’ too real".
Does anyone really write like that? This doesn't seem like racism.
It’s more data and the data based probably of the next word leading to this. There isn’t a lot of intelligence in the model as such. It’s not thinking on its own that hey today I want to start working out so that I can lose my beer belly.