Hugo, AI and Censoring Inconvenient Questions
As I have mentioned before, this blog is currently using a static site generator called Hugo. I was recently browsing a changelog for Hugo 0.155 when I came across this commit. This immediately raises alarm bells for me. LLM-generated code is demonstrably low-quality and poorly maintainable, no matter how politely you ask the LLM to write good, maintainable code. As I rely on Hugo to maintain my website, I obviously do not want it to become unmaintainable slopware. But I did not just want to assume the worst - this might just have slipped in by accident, or as part of one dev playing around with the technology for a bit out of curiosity. So I asked a question: ...