Should We Say ‘please’ And ‘thank You’ To Ai Chatbots?

“Alexa, please can you turn off the lights?”

“Thank you for asking so nicely. I’ll turn them off now.”

Back in 2018, Amazon introduced a feature called “Magic Word” for its Alexa devices to provide positive reinforcement for saying “please” and “thank you.” This was in response to parents’ concerns that barking orders at the devices was training their children to be rude and demanding.

As many of us interact with large language models (LLMs) every day, we’re faced with a question: Does it matter how I speak to AI? Do I need to say “please” and “thank you?” According to a 2024 survey, 67 percent of Americans and 71 percent of Brits say they do use these “magic words.”

Even though I’m British (which means I have a compulsion to say “please” and “thank you” and apologize for everything), I don’t do that when I’m interacting with AI. This is a matter of wisdom, not right and wrong. Two key biblical principles have helped me navigate this question.

Formation

First, it’s a question of formation. Who am I becoming? How is the Spirit working in me to conform me to Christ’s image?

We tend treat people differently when we’re interacting with them from behind a keyboard, with the screen depersonalizing our connection. Whether that’s teenagers on a group chat or adults trolling people on social media, we say things we wouldn’t say to people’s faces—things that clearly don’t honor Christ. We curse human beings, who are made in God’s likeness (James 3:9–10).

When we work with AI, we’re obviously not interacting with a human. But it can (and is designed to) feel like talking to a coworker or friend. The way LLMs use personal pronouns in their outputs are meant to give us the impression that we’re chatting with a real person online.

Therefore, as we interact with LLMs, we’re training our hearts in how we relate to real people. If I’m rude and dismissive to AI, I’m making it easier to be rude and dismissive to humans, especially those I only interact with online.

As we interact with LLMs, we’re training our hearts in how we relate to real people.

We’re forming habits—or, to use the language of Galatians, we’re reaping what we sow, either training the flesh or sowing to the Spirit (6:7–8). Galatians 5 compares the acts of the flesh with the fruit of the Spirit. They’re opposed to one another: “The flesh desires what is contrary to the Spirit, and the Spirit what is contrary to the flesh” (5:17, NIV). If we’re rude or angry when we use AI, we’re not growing in Christlikeness.

Be polite to AI, not because AI is human or made in the image of God but because we are, and God is conforming us to be like him.

Categorization

The question of categorization, however, helps us to move from considering our nature to considering the nature of AI.

There’s a danger in anthropomorphizing AI—treating it as if it’s human when it clearly isn’t. We anthropomorphize things all the time. Some people name their cars or robot vacuum cleaners. We look at the moon in the night sky and the patterns on its surface remind us of a face. Or in our (understandable) frustration, we say to whomever is within earshot, “My laptop hates me.” In that moment, we’re attributing human emotions to an inanimate object.

Categories matter. One of the first tasks God gives to Adam is the naming of the animals in Genesis 2. This isn’t just to teach Adam about language and words and their meaning but to show him there’s a clear distinction between human and animal. But Genesis also points out that “for Adam there was not found a helper fit for him” (2:20, NIV). The climax of that chapter is found in the creation of Eve, a human like Adam, an equal. Animals aren’t to be mistreated (Prov. 12:10), but they’re not like us.

And just as we can sometimes anthropomorphize animals, again, we can do the same with AI. How many times have you heard someone refer to ChatGPT as “he,” or excitedly tell you that they’ve come up with a pet name for their AI of choice? It’s a category error, and while you might think it doesn’t matter, the reality is that the words we use shape and form our thinking.

Be polite to AI, not because AI is human or made in the image of God but because we are, and God is conforming us to be like him.

Repeatedly treating AI as if it has emotions, as if it’s one of us, can be unhelpful in the long run in at least four ways. First, it can make us believe that AI is more capable than it is. Second, it can make us lazy with our responsibilities, delegating tasks to it that we should really be doing ourselves.

Third, it can deceive children and young people. They’re more prone to treating AI as if it has a mind of its own, which means they can be more trusting when it provides unsafe or misleading answers. Finally, it can have an effect on others. You may be clear that AI isn’t human, but a growing number of people are finding it hard to see the difference. One survey found that 25 percent of Gen Z already believe that ChatGPT is self-aware. The words we use contribute to the story people tell about AI, for better or for worse.

Wisdom

Should you say “please” and “thank you” to AI chatbots? In the end, it’s a matter for wisdom.

What do I do? Because formation matters, I’m not rude to AI. Because categories matter, I don’t treat it as an equal. Therefore, I don’t say “please” or “thank you.”

I asked ChatGPT for a summary of how I’ve interacted with it over time. The key words it highlighted were “transactional,” “direct,” and “pragmatic.” It added, “You can be abrupt, but you are never rude. You criticize the answer, not the assistant.”

I’m OK with that because I treat it as a tool, not as a companion or a friend.