AI Writing is Offensive to People
Status: Solved, 17 May 2026
Problem statement:
How can we make the cost of AI generated text legible?
Historically, writing implied that a human being paid attention, struggled, selected, revised, and risked their name on this piece of work. Reading is one of the few ways to spend attention to make contact with another mind, and the implicit deal is that the writer spent something comparable making contact possible. AI breaks that signal.
That does not make their reaction epistemically good. Rejecting an idea because of its origin is a kind of genetic fallacy. But their fear has a rational core: AI makes low-effort, low-accountability text cheap at scale.
My initial answer to this, as I use LLMs freely, was to just mark every post as "Made with AI". But noticing how upset people were on Twitter made me realize I need to give this more thought. Honest labels were not good enough.
asking people to read ai-generated text is offensive.
— eigenrobot (@eigenrobot) May 15, 2026
this is not because ai text is intrinsically bad. rather, the author has not paid a cost to write the text himself. this cost is a credible signal he finds its communication important.
so: not paying that cost is telling
So that makes me ask how can we make the cost of AI generated text legible? Finding a solution to this will not make the people who are offended happy, but hopefully it is an acceptable compromise for now.
Conjecture: AI Writing: Work with the Garage Door Up