Yes and both are right. It’s a matter of which is working as expected and making fewer mistakes more often. And as someone using Claude Code heavily now, I would say we’re already at a point where AI wins.