Comment by resiros

I wonder why AI labs have not worked on improving the quality of the text outputs. Is this as the author claims a property of the LLMs themselves? Or is there simply not much incentive to create the best writing LLM?