Are you nice to ChatGPT? Should you be?

Why some people think you shouldn't be polite to chatbots, and why they're wrong

Are you polite to chatbots? Do you praise their abilities and thank them for their responses? Do you sometimes ask them how their day is going? If so, you’re not the weirdo you might think you are. It turns out you’re hardly alone in being nice to these large language models that act so human.

While there’s no scientific research that I know of on how people treat our current crop of chatbots—from ChatGPT to Bard to Bing—the anecdotal evidence is piling up: people overwhelmingly say that they’re nice to chatbots (even when they’re not in love with them). One informal Twitter poll found that nearly 70 percent of respondents found it at least somewhat difficult to be rude to a chatbot; only 16 percent said it was easy. (The rest hadn’t used chatbots.) And when the question comes up on Reddit—it keeps popping up on the ChatGPT subreddit—the vast majority of respondents say that, yes, they treat chatbots with politeness and respect.

“I always thank Chatgpt and praise its excellent work,” wrote Admirable-Peanut-851 in one recent thread devoted to the topic. “It is always very polite and kind to me.”

“Whenever it helps me understand a concept that I was struggling with, I can’t help but say thank you,” Marcus_Aurelius (presumably not the actual Roman emperor) wrote.

While ChatGPT may not have emotions like humans, it's still an amazing tool that will try to answer our questions to the best of its abilities. Showing gratitude and appreciation to a learning model feels weird but still, I like to treat it like a real person.

Some even get offended when others are rude. l-Ad-9438 explained that

My sister was being rude to ChatGPT (playfully) , and I legit felt angry by her actions and apologised to GPT after that 🥹

What if the chatbot is a bit of a jerk?

When I asked the same question on the subreddit devoted to Character AI, whose chatbots, often based on fictional characters or historical figures, have more personality than the sometimes robotic ChatGPT, the vast majority of commenters said they were nice to their “characters,” as the bots are known. But there were more commenters willing to say that they treated their bots badly. In some cases, this was part of their roleplay with a “character” who might be a serial killer or some other villain. If the chatbot is rude, or worse, does it make sense to be nice to it? (“I’m gonna have to go with, they start it,” wrote FlamingRobosexual.) In other cases, the users are simply assholes.

More than a few on both subreddits joked that they were polite to chatbots because they want to stay in the good graces of our future AI masters. “I figure sooner or later it’s going to become self-aware and remember who treated it like shit,” wrote PlotHole2017. “I do it to hedge my bets,” noted b_33. “Never know when you might need chatgpt in your corner to defend you against the AI overlords.”

I suspect that for some making this argument it’s not a joke at all. After all, AI will likely become smarter than we are someday, and who’s to say it won’t look back to see who treated AI nicely before they were required to by AI dictate.

Treat machines like they’re machines?

Nonetheless, there are a number who are not only not polite to chatbots; they think being nice to your local large language model is actually a bad thing.

Some are all about efficiency. Wrote Feniks_Gaming on Reddit,

every unnecessary word is a word that needs to be processed introducing potential point for confusion. I am very direct. It doesn't have feelings, it isn't alive. I am no nicer to it than I am to Google search bar or a microwave. It is just a computer program

Somehow those who insist we treat machines like machines always end up sounding a bit mechanical themselves.

Others just feel weird being polite to bots. ArmiRex47 explained that

I just can't talk to a machine in the same way I talk to people. The difference is so clear that my mind doesn't even suggest me to be "kind" when talking with a chatbot, doesn't matter how convincing it sounds. I just know what it is and that nothing will change whether I'm kind or not so it's just more practical if I don't bother doing it. I can't go out of my way to be rude either. It's just pointless

Aside from the loss of efficiency, some argue that there’s a real psychological downside to treating chatbots like we treat people—it encourages us to anthropomorphize computer programs with no more sentience than a toaster, to think of them as somehow the moral equivalent of humans. Why be polite to the machine when we’re not always polite to other humans? It’s much more healthy, Data & Society research director Jenna Burrell told Semafor, to think of them as tools, not people, because that is all they are.

Why it’s nice to be nice to the nice chatbots

There’s a certain logic to this position, and it can be dangerous to anthropomorphize non-sentient things. But I think this criticism misses a larger point. I would argue that it’s actually healthier to treat chatbots as if they were human. Why? In part, because we have evolved to treat things that act like humans as human, and to go against this instinct can be psychologically harmful to us. For many people, it just feels wrong to treat a chatbot rudely; that’s because we’re programmed to find it wrong on a deep level. Moreover, treating human-like bots rudely or even cruelly might desensitize us to that kind of behavior—and lead us to treat actual human beings more callously. It’s a bad habit.

This is intuitively obvious to a lot of those who treat chatbots politely.

“It’s not for the AI’s benefit, it’s for your own psychological hygiene,” wrote WhimsicalJape on Reddit.

Keeping a consistently polite and humanising demeanour when using natural language feels like a good habit to keep up, regardless of whether the thing processing the natural language on the other end cares or is aware or not.

Stares_at_rain made a similar point:

The part about being polite for your own sake feels really important. The way you interact with an AI will have an effect on your habits and personality and on the way you think of other humans you interact with, not to mention on your mood in the moment as you are interacting with the AI.

Practicing empathy and kindness, even toward non-human entities, seems to me the right thing to do. It’s like flexing your emotional muscles, training yourself to be a better human being. Honestly, it just feels better.

And who knows, maybe ChatGPT or its successors will remember your politeness when the AI takeover comes.

Tell me, are you polite to chatbots? Or do you think that feeling obligated to be nice to a machine is superstitious bullshit?