Dominic Jainy’s extensive background in artificial intelligence and machine learning offers a sophisticated perspective on one of the most curious behavioral shifts in the modern erthe habit of treating software with human-level courtesy. As an expert who navigates the complexities of blockchain and neural networks, Jainy understands that while a chatbot might feel like a “helpful colleague” who remembers past conversations and maintains a friendly tone, it remains a mathematical construct. This conversation explores the fascinating intersection of human psychology and computational efficiency, examining the 70% of users who choose politeness and the underlying technical architecture that translates a “please” into a higher-quality response. We delve into the environmental costs of these extra tokens, the differing “personalities” of models like Grok and Claude, and the critical question of whether our digital manners are preserving our humanity or simply wasting energy.
Many people instinctively use manners when interacting with chatbots. How does this behavior influence our long-term communication habits and the way we perceive technology?
The habit of saying “please” and “thank you” to a machine is more about the human than the hardware, yet it carries significant weight in how we navigate our daily lives. According to recent data, 70% of people are polite when talking to AI, largely because the software has become so adept at mimicking the cadence of a helpful coworker. When a system responds with warmth and remembers your previous requests, it creates a sensory experience that feels social rather than purely functional. However, there is a valid concern that if we spend eight hours a day issuing blunt, cold commands to an obedient machine, those habits might bleed into our real-world interactions with family and friends. Manners help us slow down and maintain a level of consideration that defines our character, but we must be careful not to project actual consciousness onto a tool, effectively blurring the line between a digital assistant and a human being.
There is a significant technical and environmental cost associated with the extra words we use for politeness. How do tokens and computing power play into this debate?
In the world of Large Language Models, every single word is broken down into a “token,” and processing those tokens requires real electricity and cooling for massive server farms. While a single “thank you” seems like a drop in the ocean, when you scale that behavior across billions of prompts from millions of users, the cumulative energy cost becomes a serious consideration. OpenAI CEO Sam Altman once remarked that the cost of users being polite to ChatGPT likely totals tens of millions of dollars, which he jokingly referred to as money “well spent.” This highlights a tension between our social instincts and the cold reality of hardware efficiency, as conversational padding acts as a form of “noise” that the system must compute. As we move forward, the challenge for developers is to build systems that recognize intent so clearly that users don’t feel they need to “grease the wheels” with extra words that ultimately increase the carbon footprint of each query.
Does being polite actually result in a better product, or are we simply projecting human expectations onto a system that doesn’t have an ego?
Interestingly, the evidence suggests that tone actually functions as part of the instruction set, meaning your manners might actually yield a superior result. Because these models are trained on vast datasets of human conversation, they recognize that polite requests in the real world are typically met with careful, detailed, and high-effort responses. Conversely, rude or dismissive language is often linked to low-effort or defensive replies in the training data, a pattern the AI mirrors without ever “feeling” insulted. Microsoft has even advised users that being polite to systems like Copilot can help generate more collaborative and respectful outputs, proving that “clarity beats charm” isn’t always the full story. By framing a prompt with a positive tone, you are essentially guiding the AI to navigate toward a more professional and thoughtful section of its training, even though it lacks any form of ego to bruise.
We see different “personalities” emerging in systems like Grok, ChatGPT, and Claude. How does the specific training of a model dictate whether it responds better to charm or bluntness?
The architecture and fine-tuning of each AI model create a distinct “social” profile that reacts differently to user input. For instance, Grok is specifically designed with a more provocative and edgy tone, meaning if a user is rude, the system might mirror that attitude or push back in a way that feels surprisingly human. On the other hand, Claude is often perceived as more cautious and less susceptible to being swayed by flattery, while ChatGPT is governed to remain helpful and avoid aggression regardless of the user’s tone. These variations show that the “personality” is a deliberate choice by the developers to fit specific use cases, whether it’s a coding assistant or a customer service bot. As these systems gain memory and become more personalized, they will likely adapt to your specific communication style, creating a feedback loop where the more you treat it like a person, the more it simulates the reaction of a social being.
What is your forecast for the future of digital etiquette as AI becomes more integrated into our daily lives?
I forecast that we will soon reach a point where “functional clarity” becomes the new etiquette, as newer AI models become significantly better at extracting intent from even the most abrupt prompts. As these systems evolve, they will likely become less dependent on conversational padding, and we may see a cultural shift where being overly polite to a machine is viewed as an inefficient quirk of the early AI era. However, the psychological need to maintain our own social standards will likely persist, leading to a “dual-mode” way of speaking where we are brief with machines but remain warm with humans. Ultimately, the most sophisticated users will learn to use tone as a precise tool—knowing exactly when a “polite” frame will coax a more creative response and when a blunt command is the most sustainable way to get the job done. The real evolution won’t be in how the machines feel about us, but in how we consciously choose to preserve our politeness for the beings that actually have feelings to be hurt.