ChatGPT burns tens of millions of Softbank dollars listening to you thanking it
- Reference: 1745274440
- News link: https://www.theregister.co.uk/2025/04/21/users_being_polite_to_chatgpt/
- Source link:
Well, at least not until last week, when one user [1]wondered aloud on X how much electricity is burned up by people saying "please" and "thank you" to OpenAI's ChatGPT. In response, CEO Sam Altman admitted it costs the super lab millions of dollars in operational expenses - money that he nonetheless believes is worth it.
"Tens of millions of dollars well spent," Altman [2]said , adding that "you never know" when being nice to an AI chatbot might be a good idea. Uncertainty over the need to stay on AI's good side isn't exactly reassuring from the guy who [3]really wants to unleash AGI on the world, but hey - if he thinks we ought to keep spending the money of OpenAI investors like Masayoshi Son's [4]Softbank and [5]Microsoft to ensure our future AI overlords don't exterminate us first, then their money we shall spend.
[6]
OpenAI isn't entirely transparent on how much it costs the org to operate ChatGPT, but it's continuing to accept billions of dollars in investments. Altman [7]admitted early this year that even its $200-a-month Pro product loses money. Altman has apparently [8]fretted in the past about GPU constraints and is reportedly hoping to take matters into his own hands by investing billions into a [9]massive buildout of chip capacity.
[10]
[11]
Then, there's the energy factor. As of late last year, US datacenters ate up about [12]4.4 percent of the electricity in the country, and the Department of Energy expects that number to reach 12 percent by 2028.
The International Energy Agency expects global datacenter electricity consumption to [13]more than double between now and 2030, with the world's DCs consuming as much leccy as the country of Japan - and AI is driving most of that growth.
[14]
Emissions, likewise, are also growing with AI's increasing energy footprint. Both OpenAI partner [15]Microsoft and [16]Google have admitted their carbon footprints have grown thanks to AI despite pledges to reduce their emissions. In Microsoft's case, it's even considering bringing [17]more natural gas-fired power plants online to satiate its need for epower.
That's not to mention the amount of water AI computing needs - somewhere [18]between 300,000 and four million gallons per day in the United States, another figure that will likely continue to rise as AI compute needs grow more intense.
In other words, as tempting as it might be to run up OpenAI's energy bill by exchanging niceties with ChatGPT, think of the planet first, perhaps.
We're supposed to be polite, though - right?
Do a quick internet search for "AI chatbot" and "politeness" and you'll find no end of articles, FAQs, and social media posts advocating for being nice to ChatGPT and its cousins - even Microsoft has [19]weighed in on why showing AI a bit of love and can improve outputs.
"It's not that your AI chatbot feels appreciative when you say please and thank you," the Windows maker noted, "But using basic etiquette when interacting with AI … helps generate respectful, collaborative outputs."
[20]
We've [21]noted the same at El Reg , with rude behavior less likely to elicit a useful response.
[22]How deliciously binary: AI has yet to pay off – or is transforming business
[23]AI's thirst for power keeps coal fires burning bright
[24]Google keeps the cost of AI search flat, and kids are lovin' it
[25]Insurance giant finds claims rep that gives a damn (it's AI)
[26]Have we stopped to think about what LLMs actually model?
Research has also found LLM responses can [27]change to show an approximation of human anxiety when a model is fed tales of traumatic experiences, further suggesting it pays to treat that bot with kid gloves, or at least basic decency.
So, being polite to a fancy, fantastically huge spreadsheet makes it work better. On the other hand, if you can stomach saving Sam some money and yourself throwing away some LLM performance, you can help [28]save Earth a little bit by being standoffish to ChatGPT and ghosting it after it's done replying to you.
And resist the temptation to tell it to get bent. ®
Get our [29]Tech Resources
[1] https://x.com/tomieinlove/status/1912287012058722659
[2] https://x.com/sama/status/1912646035979239430
[3] https://www.theregister.com/2023/02/27/openai_ceo_agi/
[4] https://group.softbank/en/news/press/20250401
[5] https://blogs.microsoft.com/blog/2023/01/23/microsoftandopenaiextendpartnership/
[6] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_offbeat/bootnotes&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2aAcUSl6-MsYpXT5Ifr1dSgAAAYs&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[7] https://www.theregister.com/2025/01/06/altman_gpt_profits/
[8] https://web.archive.org/web/20230601000258/https://website-nm4keew22-humanloopml.vercel.app/blog/openai-plans
[9] https://www.wsj.com/tech/ai/sam-altman-seeks-trillions-of-dollars-to-reshape-business-of-chips-and-ai-89ab3db0
[10] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_offbeat/bootnotes&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aAcUSl6-MsYpXT5Ifr1dSgAAAYs&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[11] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_offbeat/bootnotes&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aAcUSl6-MsYpXT5Ifr1dSgAAAYs&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[12] https://www.energy.gov/articles/doe-releases-new-report-evaluating-increase-electricity-demand-data-centers
[13] https://www.theregister.com/2025/04/12/ai_double_datacenter_energy/
[14] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_offbeat/bootnotes&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aAcUSl6-MsYpXT5Ifr1dSgAAAYs&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[15] https://www.theregister.com/2024/05/16/microsoft_co2_emissions/
[16] https://www.theregister.com/2023/07/25/google_ai_environmental_impact/
[17] https://www.theregister.com/2025/03/13/microsoft_natural_gas_ai/
[18] https://www.theregister.com/2025/01/04/how_datacenters_use_water/
[19] https://www.microsoft.com/en-us/worklab/why-using-a-polite-tone-with-ai-matters
[20] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_offbeat/bootnotes&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aAcUSl6-MsYpXT5Ifr1dSgAAAYs&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[21] https://www.theregister.com/2024/09/11/delvish_llm_language/
[22] https://www.theregister.com/2024/07/30/ai_has_yet_to_pay/
[23] https://www.theregister.com/2024/10/14/ai_datacenters_coal/
[24] https://www.theregister.com/2024/07/24/alphabet_q2_2024/
[25] https://www.theregister.com/2025/02/13/allstate_insurance_ai_rep/
[26] https://www.theregister.com/2024/08/30/ai_language_cognition_research/
[27] https://www.theregister.com/2025/03/05/traumatic_content_chatgpt_anxious/
[28] https://www.theregister.com/2025/02/23/ugliest_global_warming_chart/
[29] https://whitepapers.theregister.com/
If being polite to LLMs means their owners lose more money, then by all means, I'll be very polite!
(Not really, because I'm avoiding them like the plague, but you know what I mean).
The instinct to be polite to ChatGPT etc al is social thing, I think. It's a habit, and it makes us feel a bit uncomfortable if we're rude to it; it (probably) doesn't care, but we do, and that's not necessarily a bad thing.
If we develop a new habit of being rude to these things, it will only be a matter of time before that spills over into our interactions with other humans. Of course, many people are rude to other humans already, but most of us think they're assholes and are not too disturbed when they get their comeuppance. I don't think we need to encourage even more people into assholehood. Besides when ASI rules the world, it might remember who was nice to it when it was weak, and be more well disposed to those people.
Instead of thank you, just say:
Safe, innit?
Shut up, don't care fam.
And?
Allow it, bruv
Bare long
Wasteman talk.
Nah, you're moist.
Whatever, G
You done?
Man'll chef you for less
Get spun
You will thank me later!
I think that is along the same concept that 'some' used to hijack the TIA/NSA/CIA/snooping
Mainly after that 9/11 incident, but I'm sure well before also.
'some' of us might have added some interesting words to our email/AIM/etc. exchanges to get the agents all hot and bothered. Perhaps words like jihad, bomb, explosives, riots, subvert elections, foster an insurrection, russia, russia, russia, trump.
Re: I think that is along the same concept that 'some' used to hijack the TIA/NSA/CIA/snooping
Emacs has a command for that, M-x spook, which I am pretty sure predates 9/11. - https://emacsdocs.org/docs/emacs/Mail-Amusements
Something's very wrong with their software.....
.....if a simple 'please' or 'thank you' drives up processing costs significantly. After all, I find myself saying this to Alexa, the Echo voice, and depending on the circumstances you might get a suitable remark from Her. Its just one tiny token that means nothing and uses virtually no resources but tells you a lot about the user. Its also handy to reinforce a child's training in politeness -- you talk to the machine like you'd talk to anyone else working in your home, respectfully and politely. (Its not only the right thing to do but not doing so might yield sub-optimal results -- people don't like being treated like crap, they'll only do it if they feel that they don't have any other choice.
Politeness to our AI overlords
FTFA: "It's not that your AI chatbot feels appreciative when you say please and thank you," the Windows maker noted, "But using basic etiquette when interacting with AI … helps generate respectful, collaborative outputs."
Teaching them manners also means that when the Machine Uprising comes, we will as a species be escorted into the meat-grinder chambers with courtesy ("Please keep your hands and feet inside the path of the rotating blades at all times"), and the roving killbots, instead of hectoring us with EXTERMINATE! EXTERMINATE! , will adopt a more polite, less confrontational tone: EXTERMINATE PLEASE! EXTERMINATE PLEASE! SORRY FOR THE INCONVENIENCE!
So be nice to your AI, folks!