In the case of software specifically, because agentic coding is so ubiquitous, I don’t think the benchmark of slop should be AI’s involvement. It ought to be “Is the world better off with this existing?”