/leftypol/ - Leftist Politically Incorrect

There is no "AI" ya nincompoops. It's just dead labor feasting on living labor.

Marx called capital "dead labor" that feasts on "living labor" because capital is the result of past human labor that has been accumulated in the form of money, machinery, factories, and other means of production, no matter how "advanced" and "futuristic" those means of production get. Although these things were created by workers, they become instruments of capitalist power when privately owned and used to generate profit. In capitalism, workers must sell their labor power to survive, producing goods and services whose value exceeds the wages they receive. The capitalist takes this surplus value and reinvests it to accumulate even more capital. Thus, “dead labor” feeds on “living labor” because the wealth created by workers in the past becomes a force that controls and exploits workers in the present, continually expanding through their labor. For Marx, this expresses the central contradiction of capitalism: human labor creates wealth, yet that wealth confronts workers as an independent power that dominates them. Now take that basic Marxian principle and apply it to "AI."

There is no magic "AI." There is an emerging body of technologies built by human beings, trained on human-generated data, powered by enormous human-built computing infrastructure, and sustained by significant human-made investments of energy, water, hardware, labor, and capital (dead labor!). Calling these technologies “artificial intelligence” can obscure more than it explains. The term encourages us to imagine an independent, almost supernatural intelligence that exists apart from human knowledge and effort, when what we actually have are complex systems that identify patterns, generate predictions, and produce outputs based on statistical relationships learned from data. That data comes, directly or indirectly, from people: their writing, research, art, expertise, experience, and accumulated cultural knowledge. These systems do not emerge from a vacuum. They are built upon the intellectual labor of human civilization, and their capabilities are inseparable from the quality, diversity, and limitations of the material on which they were trained.

The quality of these systems depends on much more than the quantity of data they consume. It depends on whose knowledge is represented, whose experience is excluded, how accurately the material reflects reality, and how effectively the system has been trained to distinguish reliable information from error. A system trained on a broad mixture of knowledgeable experts, careful researchers, skilled practitioners, and diverse perspectives may be useful in many contexts. A system trained on incomplete, biased, outdated, or poorly vetted information may reproduce those weaknesses with remarkable confidence. Even high-quality training data does not guarantee high-quality answers in every situation. These technologies can make mistakes, fabricate details, miss context, and produce plausible explanations that are fundamentally wrong. They can be powerful tools, but their outputs still require judgment, verification, and often the very expertise that people are being told they no longer need.

The most dangerous misconception is the idea that this technology is a black box into which anyone can type a request and receive an infinite supply of high-quality work at virtually zero cost. The interface may look effortless, but the infrastructure behind it is not free, the resources required to operate it are not infinite, and the quality of its output is neither unlimited nor guaranteed. The apparent simplicity of asking a question conceals layers of human engineering, training, maintenance, evaluation, and physical infrastructure. More importantly, generating an answer is not the same as knowing whether the answer is correct, useful, ethical, or appropriate. A person who understands a field of expertise can use these tools to accelerate their work, identify possibilities, and explore unfamiliar territory. A person who lacks that understanding may have difficulty recognizing when the same tools have confidently led them astray. Expertise does not become irrelevant when technology advances; in many cases, it becomes more important because someone still has to evaluate what the technology produces.

This is why the growing narrative that young people should stop pursuing education, abandon difficult skills, or give up on developing expertise because "AI is going to take their jobs" is so socially destructive. It teaches people to surrender their agency at precisely the moment when they need to become more capable, adaptable, and intellectually independent. Education is not merely vocational training for a specific job that may or may not exist in twenty years. It teaches people how to reason, investigate, solve problems, communicate, recognize flawed arguments, understand complex systems, and make responsible decisions. Learning mathematics, engineering, medicine, history, writing, art, or a skilled trade develops capacities that cannot be reduced to the production of a finished document or the completion of an isolated task. If society persuades an entire generation that learning is pointless because a machine will do everything for them, it risks creating a population increasingly dependent on systems it does not understand and increasingly unable to challenge their mistakes.

There are also serious economic and social consequences to normalizing this message. If workers come to believe that their skills are destined to become worthless, they may be less inclined to invest in education, negotiate for better working conditions, or organize to ensure that technological gains are shared fairly. Employers may use the threat of automation to suppress wages, reduce staffing, intensify workloads, or justify replacing experienced workers before they have demonstrated that the technology can reliably perform the work. Educational institutions may feel pressured to prioritize immediate technological trends over the deeper development of human capabilities. And individuals who are already anxious about their economic future may experience greater insecurity, alienation, and hopelessness. These outcomes are not inevitable consequences of the technology itself; they are consequences of how societies choose to deploy it and how people are taught to understand their own future in relation to it.

We should be having a more serious conversation. The question is not whether these technologies can be useful; clearly, they can. The question is how to use them without confusing automation with understanding, statistical prediction with wisdom, or convenience with genuine competence. We should teach people how these systems work, where their limitations lie, how to verify their outputs, and how to combine their capabilities with human knowledge and judgment. We should encourage students to learn difficult things, not because technology will never change their professions, but because people who understand the underlying principles of their work are better positioned to adapt when it does. We should demand transparency about the resources these systems consume, the labor and data on which they depend, the economic interests driving their development, and the real-world consequences of their implementation.

Ultimately, the greatest risk may not be that machines become capable of doing more of the work that people once did. It may be that people are convinced they no longer need to understand the world, develop their abilities, or take responsibility for what happens within it. Human knowledge is not an obsolete input to be consumed by a machine and discarded. It is the foundation on which these technologies are built, the standard by which their outputs must be judged, and the resource society must continue to cultivate if it wants to use them wisely. We do not need to worship "AI," nor do we need to dismiss it. We need to demystify it. These are human-made technologies with real capabilities, real costs, and real limitations. Our future will depend less on whether we can make them produce more and more output, and more on whether we remain capable of understanding, evaluating, and directing what they produce.