You may have noticed a profusion of M-dashes in my writing lately. I want to be clear: this is not because I’m using AI to write for me, but because I’m using AI to transcribe my dictation.
The only words on this blog actually written by AI are the excerpts — that squidge of text you see on the front page that leads you to the actual post — and the prompts for the images that have lately begun appearing at the bottom of my posts. I’ve built an automated system that parses a post, generates a prompt, and in turn generates an image. That part is automated. That’s written by Claude or ChatGPT.
Those are the two parts of this blog that have anything to do with AI-generated text. Everything else, AI is just a tool I use to transcribe my voice.
Header image prompt (openai): Large-format still life photograph, studio lighting with one hard side-light casting a sharp shadow across a vintage Dictaphone or handheld voice recorder sitting on a pale linen surface, its red record button glowing an almost wrong, too-saturated crimson against the cool neutral ground. The recorder is centered but tilted two or three degrees off-axis, unsettling the composition just enough. The shadow it throws is twice as long as it should be, reaching toward the frame edge like something the object is trying to escape.