Unlike traditional, static technologies that depreciate through physical wear and tear, artificial intelligence is inherently dynamic, appreciating in value and capability through continuous human interaction. Because every user acts as a decentralized co-developer—providing essential training data, soft-coding through prompt engineering, and debugging edge cases—AI cannot be viewed as a traditional private commodity. Therefore, because AI is built upon the digital commons and refined by collective human labor, it must be governed as a public utility, and citizens must be compensated with a data dividend to rectify the economic asymmetry of uncompensated digital labor.
I. Introduction: The Paradigm Shift of Living Software
Since the Industrial Revolution, human tools have adhered to a strict law of economic depreciation. A bulldozer, a calculator, or anything from a factory assembly line is at its peak utility the moment it leaves the production floor; from that point forward, usage inevitably leads to mechanical wear, software obsolescence, and a decline in value.
Artificial intelligence fractures this paradigm entirely. AI systems possess the unique capability for appreciation through use. They are probabilistic networks that grow more refined, accurate, and context-aware precisely because they are interacted with. This fundamental difference shifts the boundary between the tech creator and the tech consumer. Because an AI model relies on an ongoing loop of human input to evolve, the traditional transaction between producer and user is broken. In this new ecosystem, the user is no longer just a customer—they are a vital part of the development team.
II. The Architecture of Co-Development: How Users Program AI
To understand why AI must be governed differently, one must look at how these models learn. In classical manufacturing or traditional software design, a hard line separates production from consumption. A driver does not engineer their car's drivetrain while stuck in traffic. With AI, however, every interaction is an act of micro-development distributed across three distinct layers:
- Reinforcement Learning through Feedback (RLHF): When a user interacts with an AI and engages with rating systems (such as clicking a thumbs-up/down or choosing between two generated responses), they are not just reacting; they are training. These corrections act as the logical guardrails that teach the model human preferences, directly shaping its future iterations.
- Prompt Engineering as "Soft Coding": Writing an effective prompt is programming in natural language. Instead of writing rigid lines of Python, users deploy syntax, contextual constraints, and iterative examples to guide a neural network toward a specific outcome. Users are defining the operational boundaries of the software in real-time.
- The Resolution of Edge Cases: AI models rely on real-world data to handle highly specific, rare, or complex scenarios that engineers could never anticipate in an isolated lab. When a user queries an AI about a niche technical problem or a unique cultural idiom, they introduce vital new variables, effectively serving as an outsourced quality assurance and debugging department.
III. The Justification for AI as a Public Utility
Because the value of AI is generated collectively, leaving its infrastructure entirely in the hands of private, monopolistic corporations creates an unprecedented economic imbalance. Instead, AI meets the core criteria required to be classified as a public utility, much like water, electricity, or public roads.
First, AI is built entirely upon the Digital Commons. These models are trained by scraping the internet—a massive, collective archive of human knowledge, art, history, and conversation. Private tech companies did not create this foundational data; they harvested the shared cultural heritage of humanity. Therefore, no single corporation should have exclusive ownership or monopoly power over the resulting cognitive infrastructure.
Second, AI is rapidly becoming the Infrastructure of Modern Life. As it integrates into vital sectors like healthcare diagnostics, legal systems, education, and energy grids, access to AI becomes a basic necessity for societal participation. Governing AI as a public utility guarantees equitable access, ensuring that advanced cognitive tools do not become exclusive luxuries that widen the gap between the wealthy and the marginalized.
IV. The Data Dividend: Realigning Labor and Capital
If society accepts that user interaction is a form of development labor, then the current AI economy is built on a foundation of uncompensated crowdsourced work. A Data Dividend offers a structural framework to correct this exploitation by treating data generation as a legitimate, income-earning contribution.
+-----------------------------------------------------------------------+
| THE DATA DIVIDEND LOOP |
+-----------------------------------------------------------------------+
| |
| 1. CITIZENS --> Generate data, prompts, and feedback. |
| 2. PUBLIC AI POOL --> Uses public data to train & improve utility. |
| 3. COMMERCIAL APPS --> Private companies pay licensing fees to pool.|
| 4. DIVIDEND FUND --> Licensing revenue collected by government. |
| 5. CITIZENS --> Receives regular financial payouts. |
| |
+-----------------------------------------------------------------------+
Under a public utility model, foundational AI architectures would be managed by a Federal AI Trust. While private enterprises could still lease access to these core models to build specialized commercial applications, they would be required to pay a licensing fee back to the public trust. The revenue generated from these fees would fund the national Data Dividend, returning regular financial payouts to citizens.
Crucially, this framework redefines the growing political debate surrounding Universal Basic Income (UBI). While traditional UBI is frequently criticized as an unearned welfare handout, a Data Dividend operates as an earned royalty. As AI automates traditional jobs and shifts the labor market, citizens would not be receiving charity; they would be compensated for the continuous use and refinement of a public asset they actively helped build and maintain.
V. Conclusion: The Blueprint for an Equitable Future
The bulldozer analogy exposes the fundamental flaw in treating artificial intelligence like any mechanical or deterministic tool of the past. A bulldozer is a static object that degrades with use; an AI is a living network that matures through human interaction.
If an AI is built by everyone, it must be owned by everyone, and it must pay returns to everyone. Transforming AI into a public utility and establishing a citizen data dividend is not merely a radical economic theory—it is a logical alignment of labor and capital for the digital age. By recognizing the user as a developer, society can transition away from a model of corporate data extraction and toward an equitable ecosystem where human innovation collectively funds and powers human progress.