In a surprising revelation, OpenAI CEO Sam Altman disclosed that user politeness—expressing “please” and “thank you”—has contributed to significant operational costs for the company, amounting to “tens of millions of dollars” in electricity bills. Altman’s comment, made on social media platform X (formerly Twitter), was a response to a user who speculated about the financial impact of polite exchanges with the AI chatbot.
Altman’s lighthearted yet candid response underscores an interesting aspect of AI-human interaction. Despite ChatGPT being non-sentient, many users instinctively engage in polite conversations, such as saying “thank you” or “please.” These small acts of courtesy contribute to the power consumption of the AI’s underlying infrastructure, which includes data centers that require extensive electricity not only for processing data but also for cooling the equipment.
While these additional responses might seem insignificant, the large-scale nature of AI deployment means that the cumulative effect of millions of users can result in hefty costs. Altman acknowledged the toll, but humorously referred to the expense as “well spent.” This statement has sparked an online discussion, where users joked about the potential long-term benefits of showing kindness to AI, fearing that in a future dominated by machines, politeness might make a difference.
Premium users, who pay for faster and more advanced versions of ChatGPT, might be particularly sensitive to the cost of extra phrases. Altman’s comments also serve to bring awareness to the broader environmental and financial implications of AI technologies. As OpenAI and other companies continue to scale AI services, such remarks may spark further discussions on the sustainability and energy usage of AI-driven platforms.
In conclusion, while politeness may indeed cost millions, OpenAI appears to value user experience over efficiency. Altman’s remark reflects the balance between maintaining the human touch in AI interactions and the growing costs of running AI systems globally.
