Cat-or-dog classifier promises 100% hallucination-free: it never outputs monkey

burny_tech · x · 2026-09-22

A classic ML sarcasm bit: someone claims they trained a cat-vs-dog image classifier that is "100% hallucination-free with guaranteed type-safety," because "the model always predicts cat or dog, and never outputs monkey."

The joke pokes fun at slapping "hallucination-free" marketing labels onto anything with deterministic outputs in the LLM era.

Original post →

More from Fun

Fun channel →