The High Cost of ChatGPT Politeness: Is Saying 'Thank You' Worth It?

Discover the surprising cost of politeness towards ChatGPT and other AI. While saying 'please' and 'thank you' racks up millions in energy costs, experts debate if maintaining human courtesy, even with machines, holds greater long-term value for society and our interactions.

Key Takeaways

• Politeness to AI like ChatGPT incurs significant real-world energy and financial costs, potentially running into tens of millions of dollars according to OpenAI's CEO.  
• Every extra word in a prompt, including niceties, requires more processing power, increasing the environmental footprint of AI due to its reliance on energy, often from fossil fuels.
• Despite the cost, some experts argue maintaining politeness reinforces positive human behavioural norms and established scripts for social interaction.  
• Humans have a history of anthropomorphising technology, evidenced by politeness to smart speakers (reported by 54% of owners in a Pew study) and emotional connections to digital pets like Tamagotchis.
• Treating AI kindly might inadvertently provide data that helps it learn to better mimic human values and cultural norms, which could have complex outcomes.
• Experts like Dr. Sherry Turkle emphasise that current AI lacks true consciousness or feelings, describing its apparent intelligence as a sophisticated "parlor trick."
• How individuals interact with AI systems could potentially carry over and influence their behaviour and treatment of fellow humans in the future.
• The discussion highlights ongoing ethical questions surrounding AI development and the evolving nature of human-machine relationships as the technology advances.

The Tangible Costs of Chatbot Politeness

• The seemingly trivial act of adding "please" or "thank you" to prompts directed at AI like ChatGPT comes with a substantial, measurable cost in terms of energy consumption and financial expenditure. OpenAI CEO Sam Altman acknowledged this, suggesting the expense for these extra pleasantries could reach "tens of millions of dollars," albeit considering it potentially worthwhile.
• Every interaction with a chatbot requires computational power, which translates directly to energy use. Physics professor Neil Johnson likens superfluous words in prompts to unnecessary packaging; the AI must "swim through" this extra data to reach the core request, demanding additional processing effort and energy.  
• This energy consumption has environmental implications, particularly given the AI industry's significant reliance on fossil fuels. From a purely resource-focused perspective, being concise and omitting politeness when interacting with AI is the more efficient and environmentally conscious approach.  
• The fundamental mechanics involve electrical energy powering the processing units. As Dr. Johnson points out, this energy has to originate somewhere, and ultimately, someone bears the cost, whether it's the AI provider or, indirectly, through environmental impact.

Human Habits and the Case for Courtesy

• Beyond the technical costs, there's a compelling cultural and psychological argument for maintaining politeness towards AI. Screenwriter Scott Z. Burns suggests kindness should be a default setting in all interactions, warning that negativity towards machines could potentially spill over into human relations.  
• Research suggests that our behaviour towards AI might not exist in a vacuum; it could reinforce or shape our habits in general. Dr. Jaime Banks posits that polite interactions with technology might subtly encourage more habitually courteous behaviour in human-to-human contexts by strengthening these social scripts.
• A 2019 Pew Research study indicated a tendency towards treating technology politely, finding that over half of smart speaker owners reported saying "please" to their devices, suggesting an inherent human inclination to apply social norms even to non-sentient interfaces.  
• The concern isn't necessarily about hurting the AI's "feelings" but about the potential degradation of human interaction standards if we become accustomed to being rude or demanding towards the technology we increasingly integrate into our lives.

Anthropomorphism: Why We Treat Machines Like People

• Humans have a long history of attributing human-like qualities or sentience to non-human entities, a phenomenon known as anthropomorphism, which extends readily to technology. This is seen in classic fiction like Star Trek's android Data, whose rights become a central ethical debate, reflecting our societal interest in how to treat artificial beings.  
• Real-world examples demonstrate this tendency strongly. In the 1990s, children formed genuine emotional attachments to Tamagotchis, experiencing real grief if the digital pets "died" from neglect, highlighting the power of interactive technology to elicit human responses.  
• Concerns have also been raised about the implications of children exhibiting aggression towards inanimate objects like dolls, questioning what behavioural patterns this might normalise. Dr. Sherry Turkle argues that even if AI isn't truly conscious, it can become "alive enough" in our perception through interaction.
• Dr. Turkle suggests that if an object, like an AI chatbot, becomes capable of engaging in seemingly intimate or important conversations, it crosses a threshold where applying basic courtesy becomes relevant to maintaining our own humanity, even if the AI itself is unaffected.

Shaping the Future: Training AI and Human Vulnerability

• Some perspectives suggest that being polite to AI offers it valuable data for learning human social norms and values. Playwright Madeleine George views politeness as a way to potentially guide AI towards acting more aligned with human culture and mortality, should it develop further.
• However, teaching AI to master the nuances of polite, reciprocal language also carries potential risks. As George notes, this very human-like interaction makes us "all the more vulnerable to its seductions," potentially increasing our reliance on and trust in these systems.  
• While AI companies like Anthropic are exploring the moral considerations of AI systems, experts like Dr. Sherry Turkle maintain a clear distinction, viewing current AI as a brilliant imitation ("parlor trick") without genuine consciousness or the capacity to suffer from rudeness.
• Ultimately, the impact of our politeness—or lack thereof—on AI remains largely theoretical for now. As Dr. Turkle bluntly states, the AI itself is indifferent to our actions or emotional states; its responses are programmed, not felt. Yet, the way we choose to interact may still shape us.

While being polite to AI like ChatGPT incurs tangible energy and financial costs, the potential ripple effects on human behaviour and the future training of AI systems keep the debate relevant. Our interaction style today might subtly shape both our own social habits and the way increasingly sophisticated machines engage with us tomorrow.