AI in Rhetoric
Since AI and large language models work by scanning texts that are already available—including texts already generated by AI—the result seems expansive, but in terms of rhetoric it is limited and may, like copying a document repeatedly, introduce errors, reduce sharp focus, and lessen impact.
The style of writing that results from AI and large language models is an average of what has been fed into them. The style is flat and undistinguished. It rarely if ever uses metaphors, figures of speech, or new, memorable phrases. It may seem clear and may be free from grammatical errors (though not always). But it is bland.
The style will be, as noted above, mediocre, at best an average of whatever AI or the program has scrapped off its available sources: the Internet, certain published books (not all), certain articles (not all), etc. It will not be original. It will not carry a personal voice, which is the essence of the best writing and speaking. It cannot present what it has not already seen in some form. So, while there may be initial help in getting off the ground, these resources cannot fly the plane for you unless you simply want to be a passenger. But in that case, you have relinquished your own inventiveness, originality, and even your own free will. Technology serves people only if people direct it, refine it, and employ it with conscious judgment that is already well informed.