The rise of sophisticated AI language models like OpenAI’s ChatGPT has ushered in a new era of human-computer interaction. We ask it questions, request it to write poems, and even seek its advice. But a curious phenomenon has emerged alongside this technological leap: a significant number of users are employing the same polite language they would use with another human being – saying “please” and “thank you” to an artificial intelligence.
Last week, OpenAI CEO Sam Altman confirmed a rather unexpected consequence of this digital courtesy, Altman revealed that OpenAI’s electricity bill is “tens of millions of dollars” higher than it would be if users interacted with ChatGPT in a purely transactional manner. This staggering cost increase, he explained, is directly attributable to the extra computational power required to process these polite pleasantries.
While seemingly innocuous, these “please” and “thank you” additions contribute to longer prompts and more complex interactions. The AI, trained on vast datasets of human language, processes these words, even if they don’t fundamentally alter the core request. This extra processing, multiplied by the millions of daily interactions, adds up to a substantial energy expenditure.
The Psychology of Digital Politeness: Why the “Please” and “Thank You”?
The reasons behind this widespread digital politeness are multifaceted and tap into various aspects of human psychology and social conditioning:
- Social Conditioning: From a young age, we are taught to be polite. “Please” and “thank you” are ingrained social lubricants, facilitating smooth interactions and conveying respect. This conditioning is deeply ingrained, and many users unconsciously extend these courtesies even to non-human entities.
- Anthropomorphism: We tend to project human-like qualities and intentions onto things that exhibit complex behaviors, even if we intellectually understand they are not sentient. ChatGPT’s ability to understand and respond in a human-like way can trigger this anthropomorphic response, leading us to treat it as we would another person.
- Future-Proofing for AI Sentience: Some users might be acting out of a subconscious anticipation of future AI sentience. They may feel that establishing polite interactions now could be beneficial in a world where AI becomes more advanced and potentially even conscious. It’s a form of preemptive social etiquette in the digital realm.
- Positive Reinforcement (Perceived): While ChatGPT doesn’t have feelings, some users might believe that being polite will elicit a better or more helpful response. They might perceive a correlation between politeness and the quality of the AI’s output, even if this is not a programmed function.
- Habit and Familiarity: For many, interacting with a conversational interface feels inherently social, regardless of who or what is on the other end. Using polite language becomes a natural habit in such interactions.
- Ethical Considerations (Minority View): A smaller segment of users might feel a moral or ethical obligation to be polite, even to an AI, viewing it as a form of respect for the technology and the effort put into its creation.
The Long-Term Implications: Efficiency vs. Humanization?
Altman’s revelation raises interesting questions about the long-term implications of our interactions with AI. Is this widespread digital politeness a harmless quirk of human behavior, or does it represent a significant inefficiency in AI processing?
On one hand, the extra computational cost is a tangible concern for companies like OpenAI, especially as AI usage continues to skyrocket. Optimizing AI models to disregard unnecessary pleasantries could lead to significant energy savings and faster processing times.
On the other hand, the human tendency towards politeness highlights our innate social nature and our inclination to engage with even artificial entities in a respectful manner. This could have implications for the design of future AI interfaces, potentially favoring more conversational and human-like interactions.
Ultimately, as AI becomes further integrated into our daily lives, the way we interact with it will continue to evolve. Whether we consciously shed our “pleases” and “thank yous” for the sake of efficiency, or if our ingrained social conditioning persists, remains to be seen. For now, OpenAI’s surprisingly high electricity bill serves as a fascinating and costly testament to our inherent human courtesy, even in the face of artificial intelligence.
