Comment on Apple just proved AI "reasoning" models like Claude, DeepSeek-R1, and o3-mini don't actually reason at all.
We also reward people who can memorize and regurgitate even if they don’t understand what they are doing.