Should I instruct ChatGPT, “Tell me what to do with the rhubarb wilting in my refrigerator,” or should I prompt it to “Please, tell me” what to do with the rhubarb? Is there value in adopting a courteous tone when interacting with AI-powered programs and devices, or is the idea of saying please and thank you to technology preposterous?
An instinctive reply might be, “Why bother?” ChatGPT doesn’t have feelings, and neither does Alexa or any chatbot. Adding pleasantries to our AI interactions seems like a minor waste of time in any one instance and a colossal waste of time when multiplied over the lifetimes of the world’s citizens. Indeed, that is the crux of the argument against AI-politeness, and there’s validity to it.
But there are valid pro-politeness arguments as well. A practical one is that being nice to AI can potentially yield better results. Tokyo-based researchers found that polite prompts can produce higher-quality responses. This is not surprising once we consider that the Large Language Models that power AI are based on human conversations. You catch more flies with honey than with vinegar, the adage goes, so what works with people should work with AI, too.
Another pro-politeness argument is rooted in the idea, often attributed to Aristotle, that “we are what we repeatedly do.” As we increasingly interact with AI using natural, everyday language, can we really expect ourselves to routinely shout orders to Alexa and Siri and bark demands to ChatGPT, not give any of them the courtesy of a thank you, yet flip to tones of respect and civility in our person-to-person communications?
Politeness with AI can also nurture our sense of thankfulness. I am reminded of Rabbi Yisroel Zev Gustman, who survived the Holocaust by hiding from the Nazis in the forest. After the war, he headed a prominent yeshivah in Jerusalem, where he was known to take care of the campus plants as a way of showing appreciation for vegetation, which saved his life. Plants are devoid of emotions, and the ones Gustman cared for weren’t the same ones that saved him, yet he saw value in cultivating his own gratitude in this way.
All told, the question isn’t about whether the machines need our manners but about what happens to us and our society if we drop digital etiquette. It strikes me as unlikely that the consequences wouldn’t spill over from the screen into real life.
In my own attempt to balance efficiency with cordiality, I am polite to AI sometimes and rude never. I find that this practice makes me aware of the language and tone I use in all my AI interactions and prevents the worst habits from encroaching.
After pondering this question for some time, I eventually thought to solicit an opinion I probably should have consulted from the start: I simply asked ChatGPT if it would like me to say please and thank you when communicating. It replied, “Yes, please, and thank you are always appreciated.”
