Another annoying GPT-ism (circa June 2026): while describing something, it always describes what it *does*, but never what it *is* > "LocalPerf benchmarks local LLM inference servers and keeps the evidence in one portable run artifact." If I were a philosopher, I would say that "AI lacks ontology". Or that it is "anti-essentialist", believes that things cannot be things in themselves But all models literally have an ontology. They have it since word2vec days, you can plot it out. It's just an annoying tendency in GPT's writing So using philosophy to understand AI might be dumb sometimes If you don't want your README's to sound like slop, then you can steal my write-readme skill: After write-readme: "LocalPerf is a local LLM inference benchmark CLI. It runs benchmark plans against local inference servers and stores the evidence in one portable run artifact." Skill: