Some AI models get more accurate at maths if you ask them to respond as if they are a Star Trek character, ML engineers say::Researchers asking a chatbot to optimize its own prompts found it was best at solving grade-school math when acting like it was on Star Trek.
It is only logical that an algorithm trained on the ways of a Vulcan, is precise and accurate in it operation and communication. Vastly more fascinating are the result when you ask it to behave like a human.