Obviously they aren't actually people so there are many low hanging differences. But consider this: Using words like please and thank you get better results o

Obviously they aren't actually people so there are many low hanging differences. But consider this: Using words like please and thank you get better results out of LLMs. This is completely counterintuitive if you treat LLMs like any other machine, because no other machine behaves like that. But it's very intuitive if you approach them with thinking informed by human psychology.