Simon Willison Surfaces Tim Schilling Principle: 'If You Don't Understand the LLM Output, Your Use Hurts the Project'

Simon Willison's Weblog·medium signal

Simon Willison published a quote from Tim Schilling establishing a clear standard for when LLM-assisted development becomes harmful: if you don't understand the ticket, the solution, or the PR review feedback, then using an LLM in that context is actively damaging rather than helpful. This is Willison's third responsible-use principle surfaced in one week alongside Django's LLM contribution policy and his own agentic engineering anti-patterns guide — building a coherent framework that moves the conversation from 'how to use AI tools' to 'when you must not.' The principle is directly applicable as a gate in any AI-assisted code review workflow.