In the not-so-distant past, my writing process was fairly straightforward: Sit at my computer, pull up my notes and transcripts, outline, white-knuckle my way through a draft, reread, revise, and submit the piece to an editor.
AI has blown this up. Now there are a dizzying number of ways to begin and proceed: Do I want AI to mine my Slack and email for story ideas? Interview me as I run errands? Generate multiple outlines? Propose 10 new hooks? Review a draft against my publication's editorial standards? At any point, I can invoke custom skills that approximate individual editors’ preferences, tighten the piece, or introduce more tension. And on and on.
These tools help me write a good draft faster. They catch gaps in my explanations and raise questions I’ve overlooked. They also give me a remarkable number of decisions to make at every stage of the process. The feedback is often reasonable, which makes it overwhelming. A suggestion might improve the pacing but cut crucial context, or make the piece more accessible but lose technical details that show how the work was done. Writing can’t satisfy every priority equally, but AI can keep making a persuasive case for each one.
I suspect managing this decision fatigue will become part of making great work. As AI expands not just what we can accomplish but the ways we can do it, we’ll need to build processes that keep us from endlessly reconsidering our options but leave room to try a different approach when the work calls for it (or a new model is released). This is a skill in its own right.
An important component to closing the loop is knowing when to bring in someone whose judgment we trust. When I hand a draft to a human editor, I feel physical relief—their feedback lets me commit to a direction and pour my energy into making that version, not the endless, shimmering alternatives, as good as it can be.
