I just read Chris Kennedy’s ‘The AI Crutch’ where he admits that he relies on AI to ‘sand the rough edges’ of his writing. He goes on to say, “That may be one of the more interesting unintended consequences of AI. We spend a lot of time asking whether it will make us better writers, worse writers, faster workers or lazier thinkers. The subtler question is whether it makes us less comfortable deciding that something is already good.”
He linked to Dean Shareski’s post, ‘Delight #39 Spelling Errors’. Dean is more comfortable with the rough edges, and he states, “When I find a spelling error now in an email I’m kind of delighted. It means there’s a real human writing this. A human who was busy, in a rush, attending to other human things and clicked send before they fully read everything. I can’t ever remember being bothered by this but only delighted.”
I have enjoyed reading the writing of both Chris and Dean for about 20 years. I know their styles, and while Chris will use AI (and other people) as editors, and Dean is more like me, writing and posting with rougher edges, both of them have a voice and style that I enjoy… and more importantly, they share perspectives and insights that help me learn and grow.
I’m not remotely critical of their writing strategies. I love their writing styles. I don’t care if Chris improves a paragraph with the help of AI. I’d never think critically of Dean if I saw a typo or grammatical error in his writing because he didn’t use AI. Both of them have a distinct voice that I want to read.
I’m more like Dean. In a post like this one, I’ll publish with zero AI assistance. But occasionally I’ll write a paragraph, rewrite it, get frustrated with my inability to sound clear or succinct, and I’ll throw the paragraph into the LLM I’m playing with, (I only have free versions now so I bounce around). Unlike Chris, I won’t disclose my AI use at the end of my post. I also use it for definitions that I share and again, don’t always quote or identify the use of AI.
I’m not trying to be deceptive, in the past, I would do the same with a dictionary and not necessarily state where that information came from. I’m not writing academic papers, and I’m not being graded. Although sometimes I’m being judged.
I remember an example on my Pair-a-Dimes blog. I was writing about working with students on a project and a student made an error of about 100 years and I said she was off by a millennia. A commenter not only corrected me, but questioned whether I should even be a teacher if I’m going to make a mistake like that. In a subsequent comment I thanked him for correcting me.
I see the appeal of using AI as an editor. I probably would do so if I was writing for someone else, or if I was being paid, or had a specific audience where minor errors would matter. But I’m outputting a blog post every day, and I already find it hard to find time for longer posts like this one. Adding an AI assistant or editor would do two things for me, it would stretch the time required to write, and it would compete with my personal creativity and desire to write. So, for now, I don’t use much (if any) AI.
Besides, I break a lot of writing rules.
I’ll write a single sentence paragraph. I’ll write a one word sentence. Frequently. I absolutely love to use ellipsis… way more than they are required. An AI editor would have a field day correcting my writing idiosyncrasies… and sanitize my writing style. That would make me a different writer, but would my writing be better? Grammatically, yes. More enjoyable to read? I don’t know? Maybe. What it won’t make me is more likely to write every day.
So, you’ll catch the odd spelling mistake, although I tend to find those. What’s more likely is that you’ll catch a wrong word. You’ll find a ‘do’ instead of ‘so’ or heaven forbid a ‘then’ instead of ‘than’.
‘Fir’ the record, (left that in as an honest example that actually happened), I do reread my posts before publishing. However I’m notoriously bad at (or rather good at) seeing the word I think I wrote and skipping over the errors obliviously. If you want to judge me for that, it’s your prerogative, but do me a favour… please tell me. Despite everything I just said, it still bugs me that my writing has errors. I appreciate and never feel slighted by someone helping me out with a correction.
~
Ps. I use AI a lot to create the images that go with my blog posts. Again, I don’t usually state this anywhere, but I try to add a consistent style to them, and believe the use is obvious enough that it doesn’t need stating. If it was a human artist I’d feel obligated to give credit, but I feel no such obligation with AI’s trained on humans and not giving them credit.
This was my prompt I put into Bing’s Copilot for this post’s image: “Draw a stylized, abstract, watercolour of a piece of writing with many errors being corrected like you would see on a term paper of a C+ student. However, make it such that you can’t actually read any of the writing, the words should be a little bit blurry and impossible to read, but the corrections should look quite clear in red ink.” … and I quickly edited out the ‘C+’ grade it added in the top right corner.
