I think this is correct in the general sense.

In the IBM-specific sense, the most unintentionally hilarious LLM interaction I have ever had was asking Granite’s 4.2 30B reasoning variant the car wash question.

Reams and reams of corporatese about why it wasn’t wrong when it was.