A February 19, 2025, collaboration between Claude AI and Max Abecassis

Abstract

This article argues that Advanced AI Systems represent a fundamental break from historical patterns of technological labor displacement, evidenced primarily by our current inability to identify meaningful new human labor opportunities—not due to limited imagination, but because the technology's defining characteristic is its comprehensive elimination of necessary human involvement. Unlike previous technological revolutions, where new roles for human labor were visible even in early stages, Advanced AI Systems leave no clear path for human economic participation. The analysis introduces two key theoretical frameworks: first, that previous technological transitions have systematically misattributed machine productivity to human capability, and second, that faith in technological continuity has acquired quasi-religious characteristics that blind us to this unprecedented discontinuity. 

Breaking the Pattern of Technological Employment

Prior to AGI/ASI level capabilities, dramatic labor displacement will begin to materialize with the introduction of "Advanced AI Systems" synergistically integrating Narrow AI or Artificial Narrow Intelligence (ANI). These systems include AI specialized for specific tasks or domains, such as Large Language Models (LLMs), Computer Vision Models specialized in visual perception and analysis, Expert AI Systems designed for specific professional domains, and advanced humanoid robots incorporating vision, language, improved sensorimotor control, and enhanced dexterity and manipulation capabilities.

The historical pattern of technological unemployment and reemployment will not apply to the widespread deployment of Advanced AI Systems because they represent a qualitatively different velocity and type of technological advancement—one capable of replacing human labor across nearly all sectors of the economy through comprehensive replication of human cognitive and physical capabilities. Unlike previous technological revolutions where new machinery and systems inherently created visible chains of human involvement in their production, maintenance, and operation, advanced AI systems will in due course eliminate the need for human participation across the entire value chain, including their own advancement and reproduction.

This qualitative difference is evidenced by our current inability to articulate concrete ways the overwhelming majority of humans would meaningfully participate in such an AI economy—not due to limited imagination, but because the technology's defining characteristic is its potential to eliminate the necessity of human involvement. While historical technological transitions created new industries and roles that were readily apparent even at early stages, the opacity around future employment likely signals a genuine absence of necessary human economic roles rather than mere predictive uncertainty. While in the early stages of Advanced AI Systems deployment, there will certainly be opportunities for a very limited number of specialists, these opportunities do not overcome the widespread labor dislocation.

Previous technological revolutions manifested clear paths for human reemployment. During the Industrial Revolution, powered looms displaced hand weavers but created obvious new roles for machine operators, factory supervisors, and maintenance personnel. The rise of automobiles eliminated horse-related occupations but generated clear opportunities in manufacturing, mechanical repair, and supporting industries like gas stations and parts suppliers. Even the computer revolution, while eliminating many clerical positions, spawned entire new fields in software development, IT support, and data analysis.

The key difference with Advanced AI Systems lies in their fundamental nature as widely implementable at a relatively low cost. Unlike specialized technologies that augment or replace specific human tasks, Advanced AI Systems will be designed to match or exceed human abilities across all domains—physical, cognitive, and creative. This comprehensive capability makes Advanced AI Systems qualitatively different from all previous technological innovations.

Let's consider current professions thought to be relatively immune to automation, starting with medical practitioners. Doctors are already seeing their capabilities matched or exceeded by AI systems across multiple domains. In surgery, the progression is particularly telling: first, robotic arms became the primary contact point with patients, with human surgeons controlling them remotely. The next step, already underway, is replacing human control with AI-driven systems using advanced visual recognition and precision control—eliminating the need for human surgical expertise entirely. This progression mirrors what we're likely to see across many industries—a gradual shift from AI/robots as tools to autonomous systems that completely replace human involvement.

Beyond the operating room, AI systems are being enhanced to match or exceed human capabilities in diagnosis and treatment planning. Even the supposedly uniquely human aspect of medical care—empathy and bedside manner—is being effectively replicated by AI systems that can craft more empathetic responses than human physicians.

In the legal profession, current language models already possess the capability to provide comprehensive legal services, with the primary barrier being LLM providers' current strategic choice to avoid potential legal liabilities rather than any technological limitation. The logical endpoint of this progression will be fully automated legal proceedings where AI systems represent both plaintiffs and defendants, presenting cases to an AI judge who could process the entire body of relevant law and precedent in minutes rather than months. This is not a far-fetched scenario—it simply represents the natural optimization of legal processes once we remove the artificial constraint of human processing speeds.

The creative industries, often cited as a refuge for human employment, are similarly vulnerable. The rapid advancement in AI-generated content, from writing to visual art to music composition, demonstrates that even these supposedly human-centric domains are not immune. The entertainment industry's response, as evidenced by recent SAG-AFTRA legislation regarding AI replicas of performers, exemplifies a broader pattern of professional resistance through organized labor. This resistance, while understandable, may ultimately prove futile against the economic imperatives driving AI adoption.

Indeed, the increasing activism of unions and professional associations across various sectors represents a significant but likely temporary barrier to AI displacement. These organizations are adopting increasingly aggressive strategies to preserve human labor—demanding contractual guarantees against AI replacement, advocating for regulatory restrictions on AI deployment, and seeking to establish protected domains of exclusively human work. However, these efforts face two fundamental challenges: first, they can only delay rather than prevent the eventual economic advantages of AI adoption; second, they risk accelerating their own obsolescence by driving companies to develop fully automated alternatives that circumvent human labor entirely. The pattern is already visible in how companies respond to labor actions by investing more heavily in automation, effectively transforming temporary work stoppages into permanent labor elimination.

Task-Specific AI and Immediate Labor Displacement

Smaller AI companies are developing highly specialized, task-specific models targeting discrete job functions. These focused AI systems, while lacking general intelligence, can effectively replicate and replace specific human roles: a medical office receptionist handling appointments and insurance verification, a repair service AI managing scheduling and dispatch, or a customer service AI handling routine inquiries and complaints.

This pattern of targeted displacement is particularly powerful because it allows for incremental adoption without requiring massive infrastructure changes. Small software companies can develop these specialized models by fine-tuning existing language models and combining them with basic process automation. The investment required is relatively modest compared to developing general AI or robotics, making the business case compelling even for small and medium-sized businesses.

The advantages of AI voice-enabled customer service over current systems are substantial and immediate. Where traditional systems force customers through rigid menu trees, requiring them to listen to entire option lists and frequently repeat information, AI systems enable natural conversation where users simply state their problems in their own words. The AI can adaptively ask clarifying questions based on responses, maintaining context throughout the interaction without requiring repetition at different stages.

The most striking improvement comes in language and communication clarity. Instead of customers struggling with accent barriers during international calls, AI systems provide consistent, clear speech and perfect comprehension across multiple languages. The AI can dynamically adjust its speech patterns and vocabulary to match the user's level and style, eliminating the communication friction that often characterizes current support experiences.

The potential impact on government bureaucracies is especially compelling. Current government service interactions often involve long wait times, multiple transfers between departments, repetitive form-filling, and frequent confusion about proper procedures. An AI system could transform this experience by providing instant, accurate information about complex regulations, guiding citizens through required documentation, and ensuring consistent interpretation of rules across all interactions.

The scope of potential AI automation in government administrative roles is staggering. Approximately 2.7 million government employees across federal, state, and local levels are currently engaged in administrative, paper-processing, and customer service roles that could be substantially automated by AI systems. A conservative estimate suggests another 1-1.5 million government positions involve significant administrative components that could be automated, bringing the total potential impact to around 4 million positions, representing approximately 15% of total government employment.

The Historical Misattribution of Productivity Gains

Arguments for the resiliency of human labor in the face of technological progress are often associated with the idea that human productivity increases with technology. This historical narrative of increasing human productivity represents a uniquely consequential misattribution in economic thought. While this misattribution exists on a spectrum rather than as an absolute, the vast majority of what we label as human productivity improvements actually represents the productive capacity of capital equipment itself.

Consider the modern farmer with a GPS-guided tractor versus their historical counterpart with an ox-drawn plow. While the contemporary farmer has indeed developed new skills in operating software and navigation systems, these skills do not account for the massive productivity differential. The predominant productivity increase derives purely from the productive capacity of the capital equipment itself.

The business language surrounding technological investment systematically obscures this reality. Return on investment calculations are invariably presented in terms of "productivity improvements" when they are actually calculating a much simpler equation: (Cost of Human Labor Eliminated) - (Cost of Machine + Maintenance). This calculated ambiguity serves both psychological and social functions, maintaining the fiction that technology enhances rather than replaces human labor. Even the seemingly more honest term "labor-saving device" softens the reality by framing job elimination as mere savings.

This pattern of linguistic misdirection pervades business terminology. "Efficiency gains" really means labor reduction. "Performance enhancement" means labor reduction. "Streamlined operations" means labor reduction. "Optimizing workforce output" means labor reduction. The marketing of AI systems particularly exemplifies this tendency—vendors claim their systems will "make customer service representatives 300% more productive" rather than stating the reality: "our system will eliminate 75% of your customer service positions." This deliberate obscuring of labor reduction behind productivity language reinforces the broader pattern of misattributing capital productivity to human capability.

Modern manufacturing crystallizes this pattern. The claim that today's factory worker is "more productive" than their counterpart from fifty years ago primarily describes the output of increasingly sophisticated machinery and automation systems. While workers have developed new skills in machine operation and monitoring, these skills represent a diminishing percentage of the total productive output. Return on investment calculations systematically perpetuate this misattribution by expressing outcomes in terms of "worker productivity improvements" rather than their true nature: the replacement of human labor with machine capability and the worker's diminishing role in the actual creation of value.

This spectrum of attribution varies across different technological implementations. In software development, for instance, modern programmers might legitimately claim a higher percentage of productivity gains through their sophisticated understanding of complex development environments and architectural patterns. Yet even here, the majority of productivity improvements stem from the capabilities of modern development tools, libraries, and increasingly, AI coding assistants.

This systematic misattribution has profound implications for understanding economic inequality. The growing disparity between returns to labor and returns to capital isn't simply a policy outcome or market failure—it accurately reflects where productive capacity actually resides. What we've labeled as "human productivity improvements" has actually been the progressive transfer of productive capacity from human labor to capital equipment. The owners of capital have reaped the benefits of this misattributed productivity, while human labor's fundamental contribution has remained relatively constant or even diminished.

The emergence of AGI represents not just another step in this progression but its logical conclusion—the final transfer of all productive capacity to capital and more specifically to AGI. The current state of human-AI interaction represents a transitional phase in this progression. Consider the development of this analysis itself: while the human contribution provided the original conceptual breakthroughs and paradigm-challenging insights (like recognizing the "misattribution" and the "technological progress as religion" pattern), the rapid development, refinement, and articulation of these ideas derived predominantly from AI capabilities. This distribution of capabilities—human conceptual innovation paired with AI-driven development and refinement—characterizes our current moment in technological evolution. Yet even this balance is temporary, as AGI systems develop increasingly sophisticated capabilities for original thinking and paradigm challenging.

This framing helps explain why advanced AI systems represent a genuine discontinuity rather than just another step in technological progress. Previous transitions maintained the illusion of human centrality by attributing the productivity of capital to its human operators. The wide proliferation of advanced AI systems breaks this pattern not just by displacing human labor but by making the historical misattribution of productivity unsustainable. The economic implications are profound, as this completes the shift of all productive value to the owners of advanced AI systems, leaving human labor not just less valued, but ultimately irrelevant to the productive process.

The misattribution thesis thus provides both a clearer understanding of historical technological change and a more accurate framework for anticipating the economic implications of AGI. It suggests that conventional approaches to technological unemployment and reemployment are fundamentally misguided because they fail to recognize how productivity gains have actually functioned throughout industrial history. This understanding is crucial for developing appropriate responses to the challenges AGI presents.

Technological Progress as Religion

Notably, when initially presented with this thesis, three leading AI systems—Perplexity, Deepseek, and ChatGPT—each defaulted to defending the traditional pattern of technological unemployment and reemployment. Their responses reflexively cited historical examples of how technological disruption eventually created new jobs, suggesting that AGI would follow the same pattern. This automatic defense of technological continuity reveals how deeply embedded the notion of perpetual technological progress and adaptation has become in contemporary thought. Yet this ingrained optimism may blind us to fundamental discontinuities.

Current literature typically claims several weaknesses in the thesis that advanced AI systems break historical patterns of technological employment. These counter-arguments assert that the analysis underestimates human adaptability and creativity in finding new economic niches, that it fails to account for currently unimaginable future jobs, and that it oversimplifies the supposedly synergistic relationship between human skill and technological capability. Yet these very counter-arguments demonstrate how deeply embedded quasi-religious faith in technological continuity has become in contemporary thought.

The appeal to "human adaptability" represents circular reasoning disguised as analysis: humans will adapt because humans have always adapted. This article of faith ignores AGI's defining characteristic—its ability to fill any new niche itself through its capacity for general problem-solving and continuous self-improvement. Similarly, invoking "roles we cannot yet envision" exemplifies faith-based rather than logical thinking. The essay's central insight is that our inability to articulate future human economic roles stems not from limited imagination but from AGI's comprehensive capability replacement. Previous technological transitions created visible chains of human involvement even in their early stages—the absence of such visible paths with AGI likely signals a genuine discontinuity rather than a failure of foresight.

This pattern of faith-based thinking particularly manifests in attempts to preserve some special role for human contribution in the face of automation. The insistence on finding "collaborative relationships" between humans and machines demonstrates our reflexive need to maintain belief in essential human economic value, even as automation increasingly demonstrates its ability to operate independently. This quasi-religious belief in human indispensability is especially evident in the strategies of professional associations and unions, whose protective measures often rest more on articles of faith about human uniqueness than on sustainable economic arguments. Like many religious beliefs, this faith in technological continuity serves a comforting psychological function while resisting contradictory evidence. It allows us to avoid confronting the possibility that human economic participation might become truly optional rather than merely transformed.

AI's Siren Call

Just as the sirens of mythology lured sailors with promises of transcendent beauty only to lead them to destruction, the promises of AI-accelerated medical breakthroughs and enhanced human capabilities serve as a modern siren's call that distracts from examining how AI developments could prevent majority access to these innovations. If Advanced AI Systems can independently conduct research, design trials, and manufacture treatments without human involvement, the economic dynamics would likely concentrate benefits among those who control the systems while simultaneously eliminating healthcare jobs and reducing the general population's earning capacity.

Consider current trends in pharmaceutical research, where AI systems are already accelerating drug discovery and development. While this promises breakthrough treatments, the economic model still requires massive returns on investment, leading to drug prices that many cannot afford even with current income levels. In an Advanced AI Systems-driven economy where human labor is largely obsolete, access to these innovations would likely become even more restricted, not less.

The siren's call of AI advancements particularly misleads through its promise of democratized access to expertise. While AI systems might theoretically provide high-quality medical diagnosis or legal advice at lower costs, this ignores how the elimination of these professional career paths would reduce economic mobility and concentrate wealth among Advanced AI Systems owners. The promise of cheaper services becomes meaningless when the means to pay for them has been eliminated.

The counterarguments suggesting that Advanced AI Systems' productivity gains will materially decrease the costs of human sustenance ignore the plight of unemployed individuals without means to acquire basic necessities. In the context of an economy dominated by a few Advanced AI Systems, it may be necessary to drastically alter economic structures and institutions to increase minimal subsistence assistance, provide mandatory equitable access, and secure financing from the economically and politically powerful Advanced AI Systems that control these resources.

________________________________________________________________________

About the Authors

Max Abecassis and Claude (2025-02-14)

Max is an inventor (51 U.S. patents), innovator/entrepreneur (customplay.com), and futurist/philosopher ("Beyond the Romantic Animal" 1970). In collaborations with LLMs, principally Claude, the authors have published a sequence of essays at nissim.com investigating the feasibility of establishing one or more "Isolated Self-Sustaining Human Sanctuaries" (ISSHS).