A Response and Complement to OpenAI’s "Industrial Policy for the Intelligence Age"
Executive Summary
We welcome OpenAI’s call to design a new industrial policy for the transition to superintelligence. The proposals to democratize access, share prosperity through a Public Wealth Fund, and build adaptive safety nets provide a robust economic foundation for the Intelligence Age.
However, historical and anthropological research demonstrates that economic policy alone cannot manage massive technological leaps. Whenever human societies have experienced sudden surges in productivity—from the Agricultural Revolution to the Industrial and Digital Ages—the surplus of time and resources has rarely led to widespread leisure. Instead, societies have historically invented new layers of administrative, performative, and compliance-driven labour to maintain social hierarchies and satisfy a deep-seated cultural belief in the moral necessity of continuous toil.
In 1930, economist John Maynard Keynes famously predicted that a century of technological advancement would result in a 15-hour workweek.[1] Keynes was right about the technology, but wrong about human sociology. As anthropologist David Graeber documented, rather than liberating us, the system created "bullshit jobs"—whole sectors of administrative and managerial labour designed primarily to keep people occupied.[2]
Pillar I: Structural Time Reclamation (Escaping the "Busyness" Trap)
Because employers bought time rather than output, any idle time was viewed as theft. Without structural intervention, the integration of AGI will not shorten the workweek; it will simply replace manual or cognitive tasks with performative management, endless alignment meetings, and digital presenteeism.
- Decoupling Status from Headcount: Modern corporate governance often mirrors ancient Roman patronage, where a wealthy citizen’s prestige was measured by the number of clients and dependents attending him in public.[4] Today, leadership prestige is equated with human headcount, incentivizing "managerial feudalism".[5] We must incentivize organizations to measure efficiency by output and societal value, actively penalizing administrative bloat and the creation of redundant supervisory roles.
- The Right to Disconnect and Asynchronous Autonomy: Transitioning to a 32-hour workweek is insufficient if those hours are consumed by hyper-responsive digital micromanagement. Policy must protect "asynchronous autonomy," legally returning human labour to a task-oriented model. When an AI-assisted task is complete, the human worker must be culturally and legally free to reclaim their time without satisfying an outdated, industrial-era metric of digital "presence."
Pillar II: Meaningful Agency (Preventing Artificial Bureaucracy)
OpenAI envisions empowering "AI-first entrepreneurs" and transitioning workers into the "care and connection economy." While noble, history warns us of the Law of Administrative Bloat. In 1955, C. Northcote Parkinson famously argued that "work expands to fill the time available".[6] Analysing the British Admiralty, Parkinson showed that as the number of Navy ships actively declined, the number of bureaucratic officials managing them exponentially increased.
If AI generates infinite content and regulations, the bureaucratic instinct will be to hire humans to manage the surplus. Humans will inevitably be relegated to the exhausting role of "duct-tapers" and verifiers of automated systems.
- Zero-Friction Compliance: As AGI models become capable of managing regulatory complexity, we must adopt a policy of "AI-to-AI regulation." Governments should deploy AI to audit corporate AI, explicitly protecting human workers from becoming the administrative intermediaries between two machines. The goal of safety regimes must be to automate compliance, not to spawn a new, Parkinsonian industry of human "box-tickers".
- Protecting the Sanctity of Care: As displaced workers move into healthcare, education, and community building, we must fiercely protect these domains from hyper-quantification. The danger of professionalizing care is that it invites layers of measurement and oversight. Policy should prohibit the use of AI to unnecessarily track or optimize human empathy. The value of human-centered work lies in its unquantifiable nature; subjecting it to rigid algorithmic oversight strips it of its dignity.
Pillar III: The Post-Toil Social Contract (Decoupling Worth from Work)
The most profound barrier to an abundant AI future is not technological, but cultural: the entrenched belief that human worth is directly tied to economic suffering. As sociologist Max Weber famously diagnosed, the "Protestant Work Ethic" reframed continuous, gruelling labour as a moral and spiritual imperative.[7] We are terrified of an unoccupied population. OpenAI’s proposal for a Public Wealth Fund is a vital mechanism, but it must be structured to break this psychological trap.
- Unconditional Abundance Dividends: Any Public Wealth Fund must be strictly unconditional. History shows that when states provide resources to the displaced—from Roman "Bread and Circuses" to modern welfare—they often attach punitive "performative labour" requirements to satisfy the moral anxiety over idleness. To truly unlock human creativity, the dividend must be an irrevocable right of citizenship, requiring no proof of "busyness."
- Investing in the Vita Contemplativa: The ancient Greeks and Romans viewed otium (leisure) as the noble pursuit of truth and civic engagement, while business was literally defined as negotium (the negation of leisure). As philosopher Josef Pieper argued, civilization only advances when humans have true, unstructured leisure.[8] A superintelligent society must fund "meaning infrastructure". We must direct resources toward the arts, pure scientific curiosity, civic engagement, and philosophy, culturally redefining "leisure" not as laziness, but as the highest expression of human civilization.
Pillar IV: Stewardship over Compliance (Building of a Resilient Society)
OpenAI’s document rightfully identifies that a superintelligent future requires robust auditing regimes, model-containment playbooks, and incident reporting. Securing this resilience will create new jobs and offer profound new avenues for human meaning.
However, we must heed the warning of the Byzantine Empire, which intentionally designed hyper-complex, redundant institutional bureaucracies purely as a mechanism for elite social control.[9] We must prevent "AI Safety" from devolving into Byzantine corporate administrative compliance. We must transition the concept of safety from box-ticking to societal stewardship.
- The "Firehouse Model" of AI Vigilance: Historically, society has always had roles dedicated to resilience—firefighters, emergency responders, and sentinels. The defining temporal feature of these roles is readiness. The human roles required for AI oversight must operate on this model. Their value lies in their capability to act in a crisis, not in generating daily performative output. We must protect their "downtime" as a necessary, noble state of vigilance.
- The Guilds of Epistemic Trust: If the oversight of AGI is left solely to internal corporate governance, it will inevitably create a class of middle-managers whose only job is protecting corporate liability. We must cultivate independent, highly skilled professional "Guilds" to hold the mandate for OpenAI's auditing regimes. The meaning found in this work will come from a sworn civic duty to protect human reality, relying on skills that elevate human judgment over algorithmic output.
- Cultivating "Human Friction" as a Vital Skill: Efficiency is the goal of AI; therefore, efficiency can no longer be the primary goal of human labour. In a resilient society, the most vital new skill will be the intelligent application of friction. We must mandate "Human Veto" points in critical systems. The meaning of this new work lies in possessing the ethical and sociological wisdom to know when a machine's perfectly efficient logic violates human dignity or safety.
Conclusion: Moving from Efficiency to Emancipation
The transition to superintelligence is the final test of John Maynard Keynes’s 1930 prophecy: that humanity might finally solve the economic problem and face our true, permanent problem—how to use our freedom, how to occupy our leisure, and how to live wisely and agreeably.
OpenAI’s Industrial Policy provides the necessary blueprint for the engine of this new era. But an engine without a destination simply idles. By integrating these temporal and sociological policies, we can ensure that AGI does not merely make us more efficient workers in an automated bureaucracy. A society that reclaims its time needs stewards to protect that time. Let us build an economy where our machines do the work, so that we may do the living.***
[1] John Maynard Keynes, “Economic Possibilities for our Grandchildren,” in Essays in Persuasion (London: Macmillan, 1931), 358–73, originally published 1930.
[2] David Graeber, Bullshit Jobs: A Theory (New York: Simon & Schuster, 2018).
[3] E. P. Thompson, “Time, Work-Discipline, and Industrial Capitalism,” Past & Present 38, no. 1 (1967): 56–97.
[4] Richard P. Saller, Personal Patronage under the Early Empire (Cambridge: Cambridge University Press, 1982).
[5] Yanis Varoufakis, Technofeudalism: What Killed Capitalism (London: Bodley Head, 2023).
[6] C. Northcote Parkinson, “Parkinson’s Law,” The Economist, November 19, 1955; see also C. Northcote Parkinson, Parkinson’s Law, and Other Studies in Administration (Boston: Houghton Mifflin, 1957).
[7] Max Weber, The Protestant Ethic and the Spirit of Capitalism, trans. Talcott Parsons (London: Allen & Unwin, 1930), originally published 1905.
[8] Josef Pieper, Leisure: The Basis of Culture, trans. Alexander Dru (New York: Pantheon Books, 1952).
[9] Warren Treadgold, A History of the Byzantine State and Society (Stanford, CA: Stanford University Press, 1997).
Selected Bibliography and Theoretical Framework
OpenAI. (2026). Industrial policy for the intelligence age: Ideas to keep people first.Graeber, D. (2018). Bullshit jobs: A theory. Simon & Schuster.
Keynes, J. M. (1931). Economic possibilities for our grandchildren. In Essays in persuasion (pp. 358–373). Macmillan. (Original work published 1930)
Parkinson, C. N. (1955, November 19). Parkinson’s law. The Economist.
Thompson, E. P. (1967). Time, work-discipline, and industrial capitalism. Past & Present, 38(1), 56–97.
Varoufakis, Y. (2023). Technofeudalism: What killed capitalism. Bodley Head.
Pieper, J. (1952). Leisure: The basis of culture (A. Dru, Trans.). Pantheon Books.
Weber, M. (1930). The Protestant ethic and the spirit of capitalism (T. Parsons, Trans.). Allen & Unwin. (Original work published 1905)
Saller, R. P. (1982). Personal patronage under the early empire. Cambridge University Press.
Treadgold, W. (1997). A history of the Byzantine state and society. Stanford University Press.