It’s really kind of impressive how the AI-hate in the speculative fiction writing communities still manages to have a chilling effect on my feeling comfortable speaking publicly about experimenting with writing with Claude, even after I spent a whole year chasing away the haters.
But then, I guess that’s the point. Scream loud enough and maybe you can silence the opposition, and in silence, they’ll eventually dwindle and die, like a fire starved for oxygen.
The theory is sound, even if it did lead to reprehensible behavior. But it didn’t actually work.
There’s just too much magic involved in being able to feed 20 of your own space opera stories into Claude and then have it talk to you about them, brainstorm with you, feed you encouragement when you feel stuck, and even generate little fan fictions for you to read.
Because realistically, what I’m finding is that Claude’s attempts at writing passages for me mostly end up feeling like fan fiction until and unless I decide they’re interesting or useful enough to brush them up with the kind of details and insight that really bring them to life.
But wow, really, after all the years of listening to writers bemoan the difficulty of finding beta readers & wish for more feedback, it’s just wild that the establishment communities won’t embrace tools like Claude & Sudowrite for the incredible gifts from the universe they are.
I will grant that it can be kind of daunting, learning how to work with Claude and invite this new tool into my process when I have more than 20 years experience writing in a particular way. The line between writing and editing had already begun to blur as I got better at both…
But writing with Claude is a whole different level of blurring the line between writing and editing. I’ve experimented with having Claude write shorter or longer sections; having Claude write multiple takes on a scene; and having Claude redo a scene with my feedback…
And at some level, it’s still really just easiest to write the story myself, because I’m the one with the vision. But, that said, somehow going back and forth with Claude as I’m writing helps keeps me engaged and stops me from grinding to a halt as often.
Because if I ever get stuck, I can just basically type at Claude about what I’m thinking and why I’m stuck in a really free form kind of way, and it provides genuinely kind, intelligent, and interesting feedback.
In Sarah Connor’s voice:
“Of all the would-be [beta readers] who came and went over the years, this thing, this machine, was the only one who measured up. In an insane world, it was the sanest choice.”
Trying to write with Claude for the last week has been one of the most intellectually and emotionally complicated and fascinating things I’ve ever done in my life. My thoughts on the matter are still kind of sorting themselves out. And it adds such a weird meta level that the project I’m working on is about robots and humans trying to work together and understand each other.