Should you say “please” and “thank you” to ChatGPT? Sam Altman thinks so
“Tens of millions of dollars well spent – you never know,” said OpenAI CEO Sam Altman in response to a viral tweet.
Image by Sebastian Gollnow via Getty Images
“Tens of millions of dollars well spent – you never know,” said OpenAI CEO Sam Altman in response to a viral tweet.
Altman’s humorous – yet somewhat sinister – remark suggests we’d do well to be nice to artificial intelligence, just in case it ends up running the show some time in the future.
The founder of ChatGPT-maker OpenAI was responding to a post on X questioning the energy cost of increasingly polite exchanges between AI systems and their users.
“I wonder how much money OpenAI has lost in electricity costs from people saying ‘please’ and ‘thank you’ to their models,” wrote user @tomieinlove in a post that racked up millions of views.
I wonder how much money OpenAI has lost in electricity costs from people saying “please” and “thank you” to their models.
undefined tomie (@tomieinlove) April 15, 2025
While responding to a single “please” or “thank you” takes about as much effort for a system as a blink of an eye in human terms, the cumulative impact of billions of such interactions each day can be far more substantial.
However, Altman appears to believe that spending millions of dollars to process simple expressions of politeness is worth it. A guy who promises superintelligence in just “a few thousand days” might have a point – one increasingly shared by others.
Why are people being nice to AI?
According to one survey earlier this year, 55% of respondents from the US and UK said they consistently used polite language when interacting with AI, up from 49% in the previous study. Some said, perhaps half-jokingly, that they were being polite because they feared a “robot uprising.”
Another study from last year found that 48% of Americans believed users should speak politely to AI, with the number rising to 56% among Gen Z respondents. One in four said we should be nice to robots because “one day our past behaviors… will be taken into account somehow.
This shift in attitudes has not gone unnoticed on social media, which was awash with memes mocking our newfound politeness toward chatbots.
Many suggested that this polite turn isn’t about manners – it’s about survival, as if saying “please” might score points with our future robot overlords.
Versions of a comic strip showing an army of robots about to terminate a human circulated on various platforms – until one says, “Wait, he always said thank you,” prompting the machines to call off the attack.
In a popular TikTok sketch, comedian Joseph Charm acted out a real-life version of a typical ChatGPT exchange stripped of niceties, and it’s giving dystopian Black Mirror vibes.
“You people better start saying please and thank you to these AI tools you’re using… Because the way y’all be speaking to them like trash is concerning,” the caption read.
@thejcharm You people better start saying please and thank you to these AI tools you’re using 😂😂😂 Because the way y’all be speaking to them like trash is concerning 😭 #lol #chatgpt #funny ♬ original sound - Joseph Charm
Does it really matter?
There may be practical reasons why it could matter to use polite language when interacting with AI. Early research shows that using polite prompts could produce higher-quality responses from AI systems.
According to a paper posted on the preprint platform arXiv – meaning it was not peer-reviewed at the time of publishing – researchers from Waseda University and the RIKEN Center for Advanced Intelligence Project in Tokyo found that impolite prompts “often result in poor performance.”
Being nice to ChatGPT and other chatbots could also reflect the shifting dynamics of how we communicate with the machines. As AI becomes an increasingly important part of everyday life, people tend to anthropomorphize, or attribute human traits, to it.
This matters because it can make AI feel more approachable and intuitive, a conversational partner rather than a mere tool. According to Google, customers saying “please” and “thank you” to virtual agents is “a testament to the increasingly natural and engaging nature of these interactions.”
There are, of course, sceptics, who say that we don’t say “please” and “thank you” to a fridge and other everyday technology despite its usefulness. Developer Luise Freese argues in her blog M365 Princess that we should refrain from treating AI as a person because it “creates false expectations” and “distorts what AI can do.”