How and/or why do you use LLMs?
(last revised: 19/07/2026)
To me, it's a tool that is good or bad depending on the person. I used to feel very negatively about it myself, but as of now I use it to generate large swathes of boilerplate code and help me write more comfortably. All the writing you read here is mine, albeit simplified for cognitive accessibility. Without these modifications, I run the risk of confusing you and creating a worse reading experience. I have two primary explanations for why this is.
Firstly, I am naturally verbose; I nest half a dozen clauses per sentence, and this does not make for comfortable reading, even for my intended audience of naturally verbose nerds. Secondly, I have PTSD, which can interfere with the process on top of these linguistic quirks. Of my nervous system's ideas of how to handle stimuli, dissociation is the one that intrudes the most. Keeping occupied is how I prevent dissociation, as with its sibling ailment, derealisation, a disconnection from the surrounding environment instead of the self. Thankfully, doing this is not overly difficult, being autistic means a tight routine is desirable for me. I start a passage by recording on my Iphone's Voice Memos program, copy the transcription and then I exchange with a robot until I whittle it down to a more concise result, adding and removing passages along the way. I find myself motivated to write more, especially as I remove chunks of the transcription. Eventually I review it for cadence and read it out loud, where I may repeatedly rewrite passages until I am satisfied.
I do not hold my drafts in high regard, for in the words of Hemmingway, "The first draft of anything is shit." I also find using a robot is more efficient than stopping Lutz after a hard day's work and making them answer heaps of writing questions when they'd rather be getting stoned, drawing comix, and drowning people in Rollercoaster Tycoon while we caterwaul on a phone call together. They help me in many wonderful ways, but to me, writing is a process that is often best left to a single person, even if a robot acts as their teleprompter. As long as something gets to exist and I am in control, I am okay.
Where I do draw a hard line with AI, besides not lifting any text from it verbatim, is with my visual art. There is also no place for AI to fit into my process, and, in contrast to my issues writing, I am far more confident when I draw. For reference posing I use human sources, and when I get stuck on any compositions, I ask Lutz to help, as unlike writing, drawing is much easier to multitask with, and we both tend to be drawing at the same time when we have our calls. Without feeling a particular urgency to explain my personal philosophies on AI art, I do not feel threatened by it or worry about the concept of it replacing me. I am confident in the character of my work, and its ability to weather the deluge that roars past it.
The way I see it is that AI art is a transient medium that cannot embed itself into our hearts the way our hands can. I am not afraid of pattern recognition, however advanced, when we still have true human insight. If you'd like further reading, here are some articles on the subject.