If that were true I would expect to see prose that more closely resembles the "caveman" messages found in the HuggingFace attack than the overly flowery nonsens

If that were true I would expect to see prose that more closely resembles the "caveman" messages found in the HuggingFace attack than the overly flowery nonsense we see in AI blogspam.