The question of whether you should say please or thank you to ChatGPT is having something of a moment right now, and that’s made me consider my own interactions with the popular chatbot.
I’ve never really thought about it before, but on reflection I’m starting to wonder if I’m displaying sociopathic tendencies, because I never say please and thank you to ChatGPT, or am I being gaslit into thinking I’m at fault by well my well-meaning work colleagues who look at me in horror?
The news that I never say please or thank you is probably coming as a great relief to Sam Altman, CEO of OpenAI, who recently admitted that people saying please and thank you to ChatGPT are costing the company millions of dollars in processing power, not to mention the harm all that extra processing will be causing to the environment.
My reasons for no-frills approach to AI aren’t coming from a concern for OpenAI’s bottom line, or to do with environmental awareness issues: I simply can’t bring myself to be polite to machines, and also, like most of us, I’m in a hurry. I’ve got stuff to do, and it takes longer to be polite.
I like to think that this behavior stands in stark contrast to how I treat people in real life, especially my work colleagues (although you’re going to have to take my word for that).
Essentially, I’m a big ‘please’ and ‘thank you’ sayer in real life, but I don’t see ChatGPT as a person, in the same way that I don’t see Alexa as a person, or my car as a person. (Incidentally, I don’t say please and thank you to Alexa or my car, either.)
It’s not like adding a please or thank you to a ChatGPT, or Alexa, request makes any real difference to the answers you get, either. Does ChatGPT care that I’m polite? It doesn’t – you get exactly the same results from asking it to ‘find me 10 examples of people in sport who quit after not getting a pay rise’ as you do if you ask it the same question and add ‘please’ at the end. I know because I just tried it.
When you say ‘please’, ChatGPT starts its response with ‘Certainly!’, but you still get the same answers. The rest is just window dressing.
Also, I don’t feel like I’m actively being rude to ChatGPT. I’m just missing out unnecessary words that don’t make any real difference.
Our survey says…
So why do people say please and thank you to machines?
A 2024 study by Future (the publisher of TechRadar) revealed that 67% of US people are polite to AI. That’s staggeringly high.
I get that politeness is an ingrained habit that’s hard to break for some people, and that you get your answer framed in a slightly different way when you’re polite (see the ‘Certainly!’ example above), which means ChatGPT feels all warm and cuddly when you interact with it.
But beyond that, I think the logic starts to get a bit hazy. One person I know genuinely insists that they are polite to ChatGPT because one day it’s AI that will be in charge, and it will somehow remember them.
Yes, I’m sure that when the AI killer robots executing the last remains of the human race, they will pause and say, ‘Stop, this is Kevin, he always said thank you. We’ll let him live.’
Ghost in the machine
What worries me is the tacit assumption that if you’re rude to ChatGPT then you must be rude to people in everyday life as well, as if a person can only exist in one mode of being. I’d argue that if your politeness is based purely on habit, then are you really being polite? Or are you just unthinkingly acting out patterns of behaviour without any genuine emotion behind them?
I can tell the difference between people and machines, and treat each accordingly.
If ChatGPT ever (and depending on who you talk to this is either inevitable, or impossible) reaches the level where it can be said to be conscious, rather than the pattern matching algorithm we know and love today, then yes, I’ll start using please and thank you.
Until then, I’m keeping my interactions with AI as short and limited as possible, and if that helps save the planet along the way then so much the better.
