How I Reach My Weekly Writing Goals Without AI Writing for Me
Writing along the machines.
This is a kind of bottle episode (i.e.: think a Friends episode w/o Chandler or Joey, or a Rick and Morty episode centered on Squanchy).
Writing along the machines
I’ve used AI to write this newsletter, but not how you might think.
AI doesn’t write for me. Few – scant!– words generated by AI tokens have made it into a published post here. But AI does help me in my writing process.
This is not a sanctimonious take on the One True Way of using AI for creative endeavors, or about not using AI at all.
We’re in an extremely experimental phase, and things will only continue to change. AI-generated content is now on par with what humans are capable of when they’re motivated by selling, advertising and being lazy, dull and “efficient”. That stages in AI’s development is well underway.
That’s why AI doesn’t write this newsletter: because asking AI to write for me is a not-even-wrong way of framing what I’m trying to accomplish here.
Having AI write for me is much like driving my car when what I want is to go for a bike ride.
My goal isn’t production or efficiency. My goal is friction! To go to the trouble of thinking! The detective work in figuring out what the hell I’m trying to say! The quest!
I want to ride my bicycle, I want to ride it where and how I want.
So I don’t ask AI to write for me. Instead, we write alongside one another!
I ask an LLM to tell me what it knows about and organization and what the LLMs would write about it. I ask it to give me three to five takes on each subject, and then to ensemble the threads I’m most interested in.
Without fail, LLMs tell me what most people would say about that organization, the industry they’re in, the problems they strive to solve and the value they aim to create. They summarize what you can glean from doing cursory research on an organization’s website plus a few Wikipedia pages.
And so far, that’s about it. Because they can’t draw a line on the sand and stand their ground! These spineless creatures of artificial intelligence! Large Language Moderates!
They cannot help it: they love to hug that middle line, and they rehash one narrative arch ad nauseam, failing to dig up any valuable nuance – much like most media outlets, I’d say.
For example, they insist on using the Open Source Software Is The Commons metaphor to understand this particular realm of the modern world I’ve been exploring here.
Mind you, OSS as Commons is not an incorrect metaphor; it simply represents a generic and partial understanding of what open source is, and what open source can be.
AI isn’t bold, it doesn’t lean forward to peek into the unknown and the possibilities beyond. Which is terrible limiting…
…and yet: incredibly helpful! LLMs help me write by telling me what not to.
I’m not an open source software expert. But thanks to my LLM collaborations, I’ve learned to sniff out for clues, truths and boo’s. LLMs are great at pointing me in the direction of the mediocre view, the midwit take, the lazy route and the sloppy, unsavory dullness of “don’t make me think” content.
LLMs will improve and my prompting, of course, can change. This is all in flux. I’m here for it.
I can tweak my prompts so that AI starts producing new thoughts and content for me, but I just don’t want to.
And if all I wanted was to “grow” this newsletter and profit, then I bet it’d be most helpful: most people would probably want to read that more than they’d want to read what I have to say. And that’s…fine! Still, it’s not what I want.
I want to write and think my way through the mud and, hopefully, come out on the other side with a new understanding and perspective, and something witty to say that’s a pleasure for you, dear reader, to chew on.
That is the wonderful practice I’m engaged in here, the one I’m inviting you to join and come along.
In the age of agents, don’t waste your own agency. Don’t give it up so easily.
I, for one, will not. And I will continue to experiment. In the meantime, why would I want to hand off the best part to the eager machines writing alongside me?