What TFA was talking about didn't really seem like an hallucination - just a case of garbage in-garbage out. Normally there are more examples of good/correct da

What TFA was talking about didn't really seem like an hallucination - just a case of garbage in-garbage out. Normally there are more examples of good/correct data in the training set than bad, so statistically the good wins, but if it's prompted for something obscure maybe bad is all that it has got. Common coding tasks are going to be better represented in the training set and give better results.