sobkoJun 8, 2025
Manton Reece

Joanne Jang who works at OpenAI has a blog post on human-AI relationships:

"…many people say “please” and “thank you” to ChatGPT not because they’re confused about how it works, but because being kind matters to them."

I read this last night and ever since I’ve been trying to figure out why I usually type to an AI chatbot with proper spelling and punctuation, even correcting my chat text when I make a typo. It doesn’t matter, the robots don’t care. But it’s almost like if I skip that step, if I’m careless, I’ve somehow compromised all of my writing.

Some thoughts on human-AI relationships

and how we're approaching them at OpenAI

Reservoir Samples
Jun 7, 2025 at 1:45amWeb
Show threadChris VerbreeJun 7, 2025
@manton this is an interesting quote. I was just thinking about this habit last week as well. I’m not so strong on right spelling (mostly due to not able to do it). But I try to write in full sentences. I think it feels sloppy otherwise.
Show threadManton ReeceJun 7, 2025
@V_ Yeah, maybe there is something about if we’re sloppy than we’re more likely to make mistakes elsewhere too.
Show threadisaiahJun 7, 2025

@manton @V_ i always say please and thank you.

and when i open a chat i say, “Hi, i’m working on…”

some might say it doesn’t matter. but it’s hard for me to imagine that’s true.

which question is more probable to be followed by a correct answer: a question asked politely or one asked rudely?

seems like a demanding question typed carelessly is more likely to be met with sycophantic hallucination.

but more than that it make me feel good when i’m nice — even to plants and cars and whatever.

Show threadKev QuirkJun 7, 2025

@manton I’ve been using ChatGPT for a couple months now for various things. Proofreading text, sanitising and improving my PHP where needed, as I’m still very much learning, and a few other bits.

I have a subscription, as I prefer to pay my way, but I’m one of those who is always polite to it. It’s also picked up on the way I “talk” and has emulated that a little, especially my sarcasm. There’s more than a few times that it’s actually made my LOL with a quip or a pun.

Oh, and I call it Geeps. “Geeps, can you proofread this post for me please pal?” And it usually responds with something equally British.

It’s fun and (more importantly) useful.

Show threadDavid JohnsonJun 7, 2025
@manton I’m pleased that you brought this up. I’ve tried not to be polite with ChatGPT - I guess just ask my question devoid of Please or Thank you, etc, but I just can’t stop myself. Not using a polite conversational tone just doesn’t feel right to me. As you say, I do it for me, because not to do so does not feel right.
Show threadJim MitchellJun 7, 2025
@manton This is a really interesting thread… I care about being kind to humans, but personally, I’ve not used please and thank you with LLMs. I make a huge effort to be courteous and concise with my prompts, but to me, LLMs are not alive. Similarly, I don’t say please or thank you to Siri or Alexa. To me, they’re simply tools.
Show threadMichael MitchellJun 7, 2025

@manton I find myself doing this too—I just chalk it up to good manners and the desire to keep practicing politeness, especially in today’s environment. And call me crazy, but I think the AI is more helpful—or at least it seems that way—when you add a “please.” Probably just a perception thing on my part.

I do think @jim and others are right to approach AI as a tool, but for me (and probably a lot of us), it’s a generational instinct to thank people—or even things—that help us, and to ask politely. The generation that grows up knowing nothing but “AI this” or “AI that” may not even think about saying “please” or “thank you.” I just hope that doesn’t carry over into how they treat real people.