Larry M. Coleman

I recently came across a long post from July 2023 called “What AI Teaches Us About Good Writing.” Those of you who have followed me for a while probably know that by “long” I don’t mean something longer than a Twitter post, I mean something that requires an actual attention span. This one is at least a twenty-minute read, but it’s twenty minutes of your life well rewarded, and you can read it for free on Noema’s website.

It’s by Laura Hartenberger, who teaches writing at UCLA. It is a wide-ranging piece that touches on many aspects of good writing, sometimes through examples of what humans should do and sometimes through snippets of what ChatGPT creates.

She ducks and weaves through several different topics, including one on voice. Finding one’s writing voice is difficult and personal, but a critical part of what makes us unique. One of the things that stood out and resonated with me the most is this passage:

But something critical is missing from [ChatGPT’s] voice: a certain sense of connection. At its core, writing is about creating intimacy between writer and reader. It’s a relational act, not a one-sided performance, and its power is in the exchange of ideas. It’s the closest we can get to inhabiting the mind of another human, the closest to escaping our own egos.

As anyone who has spent a lot of time writing or reading (or preferably both) knows, this is absolutely true. The only thing I would add is that writing is the closest we can get to inhabiting our own mind. Not just living with it, but living in it, exploring it, and using it.

Writing a prompt for ChatGPT to plop out twenty or a hundred or three hundred pages doesn’t allow us to do that. It’s like driving a car 26.2 miles and saying you ran a marathon. You went the same distance, but you didn’t inhabit those miles: you just passed through them.

She devotes an important section to plagiarism. Unfortunately, this is probably where her analysis is weakest, but that’s not really her fault. Generative AI is so new that it isn’t just an improvement, it’s a “horses to cars” leap. The phrase “paradigm shift” got taken up by business consultants so it has lost almost all of its impact and meaning, but it really means (or at least was intended to mean) a shift so large it divides eras into a “before” time and an “after” time. We don’t have a grasp on how AI fits into the plagiarism scheme because it’s too new. This is a matter no essay can do justice to; this is a thesis- or dissertation-level topic.

Appropriately, she ends that section with a question. The question is a crucial one, however, and is one of the highlights of the entire piece:

Large language models challenge our understanding of originality and ask us to reexamine what value it adds to good writing. Is an original thought in the kernel of the idea or its phrasing? What makes a phrase original — the novelty of a word sequence, the context of its use or the readers’ perception of it? Are there any new ideas, or just new ways of saying them? Can AI generate original ideas by remixing old phrases? How original should good writing be, regardless of whether the author is human, AI or a combination?

If you want to understand how AI impacts and interacts with you as a writer, this is an excellent place to start. Hartenberger touches on many of the most important issues from the perspective of a writer seeking to be human in a coming AI-driven age. It is a half hour you won’t regret.