Can AIs actually reason, or does that involve some wishful thinking? https://www.quantamagazine.org/is-ai-reasoning-right-for-the-wrong-reasons-20260731/