#8 Society and Technology Bulletin

class=\

Industrial Revolution encoded human physical strength and time within production lines. In pre-industrial societies, people allocated their time according to nature and seasons. In agricultural communities, the rhythm of work was determined by sunrise and sunset, seasonal changes, and religious rituals. “Time” was not an abstract calculation but a flexible, cyclical rhythm woven into daily life. With industrialization, time began to be measured in hours and brought under strict control. Requirements such as workers starting production at fixed times, designated break periods, and shift systems made time a managed resource defined by economic efficiency.

Under capitalist modes of production, time became directly linked to money, giving rise to the equation “time = money.” This led to a system where workers were paid based on the hours they worked. It also created a distinction between “leisure time” and “working time.” Time outside work was increasingly viewed not as productive but as a necessary period for rest and preparation for further labor. In the factory system, precision, efficiency, and continuity became desirable norms. This fostered the internalization of time discipline among individuals. People began to see managing their own time as a virtue. In modern capitalist societies, “wasting time” came to be associated with guilt.

The 21st century is creating a similar transformation by rendering our language, gestures, and habits into marketable data. Now it is not only our bodies but also our everyday digital behaviors that are directly embedded within production-consumption processes. Our behaviors, structured as datasets, have become calculable and tradable for functional outcomes. On the other hand, their transformation into data has opened the door to manipulation techniques and attention economies. Our personal behaviors, now taking the form of tangible, controllable, designable, and quantifiable data, have ceased to be authentic and inaccessible parts of ourselves and have instead been coded as components of consumption and production systems.

This condition lies at the heart of our relationship with artificial intelligence. Our adoption of chatbots and the interactions we establish with them are thoroughly intertwined with all technical production and consumption processes. When we ask an AI bot “how are you?”, we are not merely posing a question—we are training the algorithm, expanding its repertoire of interactions, and contributing to the platform’s learning cycle. In this context, one of the most refined forms of new digital labor is the “labor of expression” we produce without even being aware of it. This labor, even when anonymized, is coded, structured, and converted into an economic output by machines. Precisely because of this functionality, it carries both production and consumption costs.

OpenAI CEO Sam Altman revealed that the politeness shown by users toward ChatGPT costs the company millions of dollars. According to Altman, users appending phrases like “please” and “thank you” at the end of search queries increase computational load and raise operational expenses.

Speaking politely to AI chatbots is a widespread phenomenon. At its simplest, this stems from human social reflexes spilling over into digital environments. The human brain, even when interacting with a machine, tends to assume there is a “counterpart” behind the text—whether human or artificial. Social etiquette automatically activates. We can describe this courtesy as the application of social norms to machines.

These discourses reveal not only the impact of linguistic expressions on production costs but also the dominant mindset within the AI industry. The prioritization of concepts such as efficiency and cost optimization reflects the industrial logic underlying this technological production process. While user experience is often relegated to the background, system priorities are shaped around outcome focus, processing speed, and energy efficiency.

This approach, in a sense, reflects the invisible factory context of the digital age. Just as workers in the Industrial Revolution had their meal and rest times restricted because they disrupted the production line, today’s digital systems imply that users must be discouraged from using polite expressions like “please” because they generate unnecessary computational load.

In both cases, human gestures are coded as inefficiencies threatening productivity; production processes are reorganized not around the human but around output. Thus, even a routine form of address directed at AI carries significant clues about the production-consumption philosophy of the digital age.

AI systems, particularly large language models, execute dozens of computational layers behind every input. Authentic human-generated data is essential to train AI systems to produce and refine synthetic data. Yet the overall cost of this process is substantial. For instance, a single user query in systems like OpenAI’s or Google’s language models is processed across tens of thousands of cores within hundreds of milliseconds. The cumulative effect of these computations translates into energy consumption, and energy translates into cost. Sam Altman’s reference to “tens of millions of dollars in server expenses” is not merely a scale estimate—it is an economic and political indicator pointing to the emergence of new “digital energy classes” in this new era.

The Industrial Revolution had visible factories: smokestacks belching steam, workshops filled with the clatter of hammers. The factories of the digital age are invisible: hundreds of thousands of servers are assembled into massive “data farms” in Norway’s glaciers or Nevada’s deserts. These data centers are modern production facilities operating 24/7. Yet this production does not yield tangible objects; it produces information, predictions, profiles, recommendations, and responses. Thus, the output of the digital age is cognitive, not physical. And every cognitive output entails measurable energy consumption in watts. Therefore, every expression has a watt equivalent.

This situation leads to the metalization of many verbal expressions, reducing them to a specific accounting economy. Now, the energy cost of expressions matters as much as their content. What matters is not what we say but how much energy it costs to say it. This mental shift moves language away from its capacity to generate meaning toward a pessimistic phase of communication defined solely by measurable computational load. In a linguistic environment where only algorithmic and symbolic relationships exist, content recedes into the background while the priority of expression shifts from “saying something” to “consuming a specific amount of energy.”

Thus, language models remove language from its role as an organic process of production and begin to re-evaluate it along axes of energy, time, and cost. This ultimately risks the disenfranchisement of language’s fundamental functions. At this point, the two sides of the intellectual scale become clear: on one side, language as an indispensable tool for solving problems; on the other, language itself becoming the source of a new problem. Completely disabling language is neither practical nor ethically viable—at least for now. Yet even imagining such a solution deepens the emerging paradox.

Therefore, short-term solutions may follow Sam Altman’s suggestion: optimizing computational load and reducing linguistic redundancies. Yet in the long term, the logical endpoint of this reasoning could be the severing of language from its essential function. Although this speculation has not yet become visible, it remains a potential risk on the horizon.