Comment on Apple just proved AI "reasoning" models like Claude, DeepSeek-R1, and o3-mini don't actually reason at all.
And that’s pretty damn useful, but obnoxious to have expectations wildly set incorrectly.