DEV Community

Philip Hern
Philip Hern

Posted on Originally published at philliant.com

who did the work?

thesis

i steer the ai, but the ai does almost all of the execution on this site. i still read the diffs and i still own what ships, but the line between my labor and the model's output keeps blurring.

context

i said in full disclosure regarding this website and my writing that the site is ai-assisted. i meant i outline, draft, and steer while the ai helps with code and maintenance. that is still true.

the website layer is now almost entirely ai-maintained on my instructions, but the same pattern has crept into content. i let agents revise old posts, fix formatting, and sweep the archive. the diffs are bigger than what i would have done by hand in a single sitting. at my day job i use ai for trivial changes too, one-line fixes i used to type without thinking. the keystrokes are the model's. the merge is mine.

argument

who is the architect and owner of this site if i steer but rarely type? my name is on it. the hugo config, ci, lint rules, and cursor skills are mostly agent work i review. when the build breaks, i often ask the ai to investigate first. responsibility stayed with me. the craft moved.

is the writing still mine? ideas i care about still start in my head. maintenance is messier. if an agent smooths prose and matches site formatting, i do not know how much of the page is still my sentence-level choice. i read the diff and revert what changes my meaning. what i accept is what is blurring the lines.

i also feel the atrophy. offloading boring work saves time but costs reps. syntax i used to write from memory, edits i used to do by feel, are now completed by prompt. i still believe you need past struggle to judge ai output. i am less sure i am still doing enough hard work to stay sharp.

tension or counterpoint

compilers and spell check never made this argument feel so personal. generative ai rewrites your voice and your repo in the same session. maybe readers only care if the post helps. i care because i do not want to become a permanent reviewer of my own work.

closing

i will keep publishing under my name and keep reading the diff. i will also keep asking who did the work, and i will keep some tasks slow on purpose. steering plus selective doing is the compromise. it is not clear anymore.

related on this site

Top comments (0)