Stop Externalizing the Cost of Your AI Use to Me

Summary

The author argues that people who send unedited or poorly edited AI-generated text, code, or research are shifting the real cost of their AI use onto whoever receives it. Messages asking others to “just skim” or “double-check” transfer the work of finding the point, checking accuracy, and deciding what matters, while an AI disclaimer does not constitute consent. The article distinguishes the use of AI itself from the failure to edit its output before sharing. It connects this pattern to broader damage in systems built around human review: open-source maintainers are receiving large volumes of plausible but costly issues and pull requests, students can submit unchecked AI solutions and use instructor feedback to generate regrade requests, and peer reviewers may face a flood of half-baked papers submitted because drafting has become cheap. The author says this asymmetry erodes willingness to provide feedback, while defensive rejection also harms people who made a genuine effort. As trust declines, the article expects more reliance on small personal networks, AI-assisted filtering, and friction such as submission fees, caps, or deposits, although these measures burden legitimate users too. The proposed response is to use AI freely while cleaning its output, agreeing on appropriate formats for feedback, and treating raw output as something requiring an apology rather than a disclaimer.