In between reading my own writing and slowly making my way through a stack of novels I’ve bought but haven’t gotten to yet, I am spending a lot of reading time on literary magazine submissions and manuscript excerpts from fellow writers in various writing groups I participate in. And once again, I will be a fiction judge for WWPH’s annual book competition, which restarts this year after a hiatus focused on anthologies and the press’ anniversary.
Like everyone reading anything these days (and writing anything), my AI spidey-senses always have to be on alert, for no one wants to admit to being engaged, enthralled or outright suckered by strings of words produced by evil AI machines. I am committed to the liteary fiction snootery and ethical considerations associated with using AI tools to generate whole ideas or passages or even full length works, but will admit to using them as basic research tools, just to save lots of time figuring out things like what the tips of shoestrings are called or asking if people who lived in 1991 used voicemail. And yes, I’ve experimented with having ChatGPT and Claude pressure-test sentences, paragraphs, character names, or help me flag plot holes in longer sections of a work. (And at some point I will probably have one of those tools clean up this short blog post, get rid of typos, and it will probably suggest a mdash or two that I will edit out, even though I am one of those who has always LOVED the damned mdash.)
But this isn’t about the ethics of it all for writers, of which there are legitimate thought issues (“stolen art.”) It’s what the cynicism and skepticism about AI does to you as a reader, especially when you come across something that feels, well, perfect. Not so much from the standpoint of artistic merit, but perfect in delivery. Perfect in its usage of adjectives, its structure and form, in following somebody’s rules and forms so well that you that you think: haven’t I read this perfect thing before? 100 times before?
Yes, we are our own LLMs. The promise of the blank page for writers is that everything you know and everything you’ve read (and what you hear editors and publishers want) is influencing everything that comes onto the page in every sentence. It’s original and unique sourced from hours and hours of lived experience: reading, writing, listening, observing, visiting, sharing, arguing, struggling, developing. Not just the good writing or the bad writing, of which there is plenty, but all the flawed writing too.
A writerly friend who has done quite well for themselves asserts that Toni Morrison’s Jazz is a deeply flawed novel, but declares that it’s still a great one. The meandering backstories, the too-chatty characters and dialogue, the cringey cliches and current trendy language to be hip, being a bit more didactic than necessary, losing the voice of a character, all make the art human. It’s what in part makes them art.
The fear for me is not that writers are using AI to burst through writer’s block, find a perfect way to craft a sentence here or there, but that they will take the body of learning that is an LLM as their gospel. If ChatGPT says it so, then it’s so.
A key tactic for me when I am going down the rabbit hole of these tools—analyze this… polish this… help me flesh out this…yes, tell me more about that—is my personal one-and-done hedge when I am “talking” to AI: if you were writing this thing, how would YOU do it? I’ve successfully weaned myself from being too tempted to take ChatGPT’s suggestions (or Claude’s, which for whatever reason feels more like a tool and less like an AI binary slop machine) and let it be nothing more than a quick set of eyes to offer a perspective in the lonely space that writing can be. I say thanks and be on my way.
Having read some recent short fiction submissions, my spidey senses did get triggered by a couple of things I am seeing. Not so much the supposed telltale formatting and construction that everyone is analyzing in business writing on LinkedIn. In fiction, it seems to be something that you feel.. let’s call it artificial sentimentalization. It’s like the AI machine is saying during it’s generative state: Now is the time for some interiority and emotion, insert emotion here.. and then all the flowery slop just starts pouring out in seconds. And I am reading like, uh oh: I am being aware that this writing is constructed as if it’s been intellectually informed to inject sentimentality, versus its earnestness genuinely making me feel? There’s some intersection here, maybe, with the famous James Baldwin quote from Notes of a Native Son that makes the rounds often in out-of-context social media posts. I’ve been toting the quote around since hearing it in some class in college, and identified it as a warning/mantra for what I wanted to achieve in my writing… and perhaps, to some degree, in my life:
“Sentimentality, the ostentatious parading of excessive and spurious emotion, is the mark of dishonesty, the inability to feel; the wet eyes of the sentimentalist betray his aversion to experience, his fear of life, his arid heart; and it is always, therefore, the signal of secret and violent inhumanity, the mask of cruelty. “
I wince at Baldwin’s assertion that it’s ALWAYS the mask of cruelty (there are no punches pulled in true Baldwin fashion), but as I read this that line that Baldwin aimed towards his human contemporaries, it sounds today like he could be writing about AI.
I hope this younger generation of writers finds their voice independent of these tools that teach them to aim toward a biased aesthetic that leads to polish perfect but boring writing, but maybe not to think and feel. That’s what I think I will say to myself as I dive into the next round of submissions. A simple set of heuristics:
Does this make me think?
Does it make me feel?
