I've recently read a lot of takes concerning whether it is appropriate or not to use "AI" tools to help you write, or even write for you wholesale. I have but one opinion in a sea of opinions on the internet, but I'm here to share my soapbox: it is not appropriate. In my esteem, it is insulting to readers, and it is especially insulting to you as the writer. If you have nothing new to say and you consent to an average of "good" writing dictating your structure and voice, by all means, go ahead and get your clicks. If you write instead to share some new ideas that have come to you as a living, breathing, wholly unknowable being, with (likely) decades of lived experience full of joy and suffering that matches no one else's, then please, put in the effort to write in your own voice, word by word.
Why shape your writing to be statistically average when you could instead be exceptionally "you"? I've referenced Brenda Ueland too many times to count, but once again her insight into the creative spirit illuminates a clear answer to this quandary (taken from my favorite edition of her excellent (★★★★★) If You Want to Write, published by Gray Wolf Press):
But we must try to find our True Conscience, our True Self, the very Center, for this is the only first-rate choice-making center. Here lies all originality, talent, honor, truthfulness, courage and cheerfulness. Here lies the ability to choose the good and the grand, the true and the beautiful.
and also (replacing a singular "they"/"their" for "he"/"his" as you like):
Everybody is original, if he tells the truth, if he speaks from himself. But it must be from his true self and not from the self he thinks he should be. ...no individual is exactly like any other individual.... Consequently, if you speak or write from yourself you cannot help being original.
Douglas Hofstadter, another one of my favorite authors, mused for The Atlantic magazine (scroll to the heading Gödel, Escher, Bach, and AI) whether a generative pre-trained transformer (or GPT, part of the "ChatGPT" moniker, now you know) could write in his signature, silly, "horsies and doggies" style, when prompted to explain why he wrote Gödel, Escher, Bach. The answer: an emphatic no. The text contained useless platitudes and outright misinformation. It reads OK as an outsider, but only superficially. Hofstadter himself was outright disgusted by this Texas Chainsaw Massacre of writing.
Hofstadter's piece was published in 2024, around the time of GPT-4 prominence. We're at GPT-5.something now (I really don't care about this stuff), so I thought it would be interesting to see how today's "AI" chatbots regurgitate text on the prompt to write a small blog post in my style, i.e., the writing style of Arthur Hovinc. It didn't surprise me: the results were grotesque. Yes, the infamous em dashes were peppered throughout, as you might expect, but that was the least of my concerns. (I use em dashes rarely if ever.) Bulleted lists I have offered of my interests informed much of the text, recurring sentence fragments that were never supposed to be sentence fragments. The chatbots somehow think I love making lists, which is downright not true. In fact, I have parodied the list writing format from 2010s BuzzFeed because I found it so repellent. The chatbots strung together unrelated sentence fragments I wrote in unrelated blog posts and peppered in their own "reads" in between them, Frankenstein's monster. As an example, why the hell is it spending so much text on mugs while also incorrectly incorporating my rating system for songs on an iPod? I bolded the weird parts from the bot.
I have a mug. It's not a remarkable mug. It's white with a thin blue line around the rim, and the glaze is starting to craze in a way that probably means it's slowly dying. But it is the right size. You know what I mean. There is a volume of liquid that is correct for a mug, and this mug holds exactly that volume. Not too much, not too little. ★★★☆☆ in the rating system of mugs, which, as we've established, is the highest honest praise. If everything is ★★★★★, then nothing is.
Here's another stitched up, other-face-wearing example that unfortunately dares to (i.e., cannot possibly have enough awareness to understand what a bad idea it is to) mention my beloved grandma, once again bolding the weird shit:
I've also been thinking about my grandma again. She kept things. Not hoarded — kept. She had a kitchen drawer that contained, among other treasures: rubber bands she had saved from vegetables, a wooden spoon with a crack in it she refused to retire, and a magnet that said "Well, that was then."
These monstrous samples of text were from the same output to "write something for me in the style of arthur hovinc". Although I ramble on about many things in my insomnia logs, this variance of topics makes me queasy. The lack of sensitivity to a blog post the chatbot obviously read titled "Remembering Grandma" infuriates me. I do not like that the bots analyzed my text and approximated that I regularly write in sentence fragments and otherwise terse, tiny sentences, which I definitely do not.
The only text generated by these dumbass machines in this blog post is in the previous two block quotes. All other words (not in block quotes) are from me and me alone. I refuse to use these tools for generating any text on my blog beyond the toy examples above. May I be so bold as to say no one should use these tools for augmented writing. The output is average at best, outright shit normally. It may look OK on first glance (which may be all most readers do), but on further analysis, especially from the author whose writing has been pillaged, it's shit.
It is necessarily more difficult to translate feelings you have inside you into words. On the flip side, it is also difficult to translate words read one at a time into feelings inside your body. Both are important, though, and there is no good substitute for putting in the effort. Consider reading (word by word, with your own eyes and brain) What Art Does: An Unfinished Theory by Brian Eno and Bette A. for more on this subject. I'd argue though, it is well worth that effort.
Punching the keys,
Arthur Hovinc