This morning, I wrote a fine, introspective, most excellent article on using AI in writing, based on something I’d seen from someone else on AI.
It was well-written, I thought, and accurate for my process - which involves using AI to review and revise, for the most part - and in the process of revision (with the AI as a sort of dialectical partner)… I realized that I was going over ground I’d already covered, in Working with the Machine.
The review was … okay. It was written well enough, but needed to be stronger, based on feedback from the review process, and that forced me to work harder - to cross-reference, to broaden the content somewhat, to find examples, and that led me to content that I’d already written. The AI didn’t find it (although it could have, if I’d used my own content via an MCP and asked it to compare it to previous posts), but the AI served as a point of friction that enabled me to find it.
There were things I had in the new post that actually did shine light into some darker corners of the process - after all, “Working with the Machine” was written five months ago as I write these words, and the passage of time changes things. But they weren’t significant enough to justify the new post without significant rework, so I “killed the darling,” as writers say, destroying stuff that was dear for the writer but wouldn’t contribute value for the reader.
The process is working. The “discussion phase” with the AI showed that I was rewriting stuff without being fully aware, and that allowed me to keep it under wraps until it can contribute actual value.
So then I wrote this, as a result, because I wanted to publish something, by golly!
Oddly enough, the process I used to kill the prior post wanted to kill this post, noting that it’s “self-aware noise” instead of “defensive noise” - because I demand that the LLMs actually criticize honestly, as opposed to just complimenting what an awesome writer I am.
It’s not wrong; I don’t like silent blogs, and I do have a tendency to ignore a specific medium for long stretches of time. This is me reading someone else’s writing, reacting to it, then reacting to my own writing.
I actually think that’s okay; this blog is a long, long, long letter to my kids, exposing my innermost thought processes (but not necessarily my innermost thoughts), so that if they ever wonder how I thought and why I thought many of the odd things they know about, they have a sort of oblique roadmap. It’s me writing to me, such that others can see, really.
And it gives me a chance to laugh at myself, too, because this process of rewriting is something I’ve suffered from for a long time: I write something, and eventually the same sort of stimulus occurs that made me write it the first go-round, so I write it from where I am today… but that means there’s a hefty dose of duplication, and my current processes help me catch it a little more than my old processes, which were apparently “fire and forget,” with no cross-referencing.