What does it mean to be a writer, when AI can write?
When a startup founder says “we used Cursor and Claude to build 90% of our codebase,” nobody thinks less of them. If anything, we’re impressed—they’re leveraging tools effectively to ship faster. The value is in what they built, not how many keystrokes they typed.
But if someone told me an album I loved was AI-generated, I’d feel differently. Something would be lost. I want to know a human felt something and translated that into sound.
So where does written content fall? I think it depends on the kind of writing.
Research synthesis sits closer to code. The value is in comprehensiveness, accuracy, organization. If AI helps produce that more efficiently, it doesn’t diminish the value. You still directed it, curated it, verified it.
Personal essays sit closer to music. If my note about building Curio for my kids was AI-written, that would feel like a betrayal. The value is in my experience, my voice.
Frameworks and arguments fall somewhere in between. The original thinking is the contribution, regardless of who typed the sentences.
How I’ve approached this platform
In general, I’ve played the role of “Producer”, driving the vision and direction, and directing dozens of workers to make it happen. Just in this case, the workers are all AI-powered.
-
The research deep dives: AI did heavy lifting on first drafts. I defined every topic, structure, and quality standard. I directed the research, reviewed the output, edited extensively, verified claims. Claude synthesized; I architected and curated.
-
The frameworks and thesis: The original thinking is mine—the argument that capacities matter more than skills, the specific eight literacies. AI helped me articulate and refine. The ideas came from years of work and experimentation.
-
Field notes and personal pieces: My experiences, my voice. AI might help me edit or tighten, but the stories and experiments are lived, not generated.
If any of it sounds like slop, I’ve failed.
How I actually work
I’m not a software engineer. But I built about 90% of this site using Cursor and Claude Code. I’ve trained a dozen specialized workers on different tasks—content strategy, research synthesis, drafting, design systems, code review. Each one has detailed instructions, voice guidelines, and quality standards.
A huge part of the learning (and honestly, the fun) has been figuring out how to orchestrate these workers together. The result is maybe 100x more output, at higher quality, than I could produce alone. That’s not an exaggeration—it’s just what’s possible now if you invest the time to build the systems.
Why share this?
Firstly, it’s the truth. Too many people are publishing AI slop under their own name, pretending it’s all their work. I think that hurts credibility.
But this is also a deep and fascinating question: what’s the human contribution when AI can do the mechanical work, and much of the thinking, better than any human? Not just for writing. For every knowledge profession. And for what we teach (and show) our kids to value.
I don’t see any simple answers. I’d rather explore it openly than pretend it doesn’t exist.
If that’s something you have thoughts about, let me know.