AI slop fatigue and the push to verify and understand code

Show HN: Sphinx is a hook that quizzes you on your AI's code before shipping. The first reply was a pained "Outsource your thinking? I'm sorry, what?" but another commenter understood the motive: you push a pile of generated code and get a nagging sense you should know it better. Elsewhere the mood was sour. In the superoptimizer thread, one commenter asked whether tech posts will now come in two flavors: things prompted into existence and ghostwritten by an LLM, or human-written deep dives. The Esp32sim reply called out "gigantic, poorly-formatted AI-slop paragraphs" in the README. An Ask HN asked for a flag on AI-generated articles, and the replies argued over detection and over whether it would breed stigma.

The counter-signal is verification. Lean now ships with verified kernels, and commenters debated what "verified" means and whether it is just checkers all the way down. Both threads ask the same thing: what do you trust when generation is nearly free?

The pattern here: output is cheap, so trust and comprehension are the scarce things. Readers now judge a project by how much human care shows in the write-up.