An AI gives a plausible reason based on the wrong clues.
Spurious reasoning is like a student guessing C, then adding fancy math. The page looks smart, but the brain took a shortcut.
It wraps a shortcut in logic. Watch for it in tests and high-risk answers.
Chain-of-thought
Spurious reasoning can hide inside a chain-of-thought.
Reasoning-model
It shows how a reasoning model can dress a shortcut as logic.
Hallucination
It can give a wrong answer a pretty reason.
Reasoning Transparency
Reasoning Transparency helps check whether the reason came after the answer.