When Productive Identity Weakens

Artificial intelligence is usually discussed as a technology of production. The central questions concern which jobs it will replace, which occupations it will transform, how rapidly productivity will rise, and who will capture the gains. These questions matter, but they remain on the surface of a deeper transition.

For much of modern society, education, capability, productive output, occupation, and social recognition have been treated as if they were meaningfully aligned. Education develops capability. Capability supports productive output. Output establishes occupational standing. Occupation supplies income, independence, status, and a recognized place in society.

This alignment has always been partial. Inherited wealth, family position, class, race, gender, and institutional privilege continue to affect who receives opportunity and whose work receives recognition. Many occupations provide income without much status, agency, or narrative of development. Yet the ideal of a society in which acquired capability supports occupational and social advancement remains one of modernity’s most important organizing presumptions.

This presumption may be called productive identity: the institutional expectation that what a person learns, what a person can independently do, what a person produces, and how that person is recognized should bear a meaningful relationship to one another.

Artificial intelligence is beginning to weaken that relationship.

A person can now generate an increasingly sophisticated essay, computer program, illustration, translation, analysis, or professional report without independently possessing all the capabilities previously implied by that output. The artifact may remain polished, useful, and economically valuable. What becomes less certain is what the artifact reveals about the person who presents it.^[1]

This is not only a problem of automation. It is a problem of signaling, formation, agency, and identity.

When output becomes less reliable as evidence of capability, institutions built around that inference must adapt. Schools can no longer assume that submitted work demonstrates mastery. Employers can no longer infer technical command as confidently from visible performance. Professional credentials become harder to interpret when execution is distributed between a person and an intelligent system. Workers may become more productive while becoming less independently capable of understanding, evaluating, or reproducing what they produce.

The resulting disruption will not be absorbed uniformly. Formally modern societies continue to operate through different informal institutions and inherited norms concerning family, duty, autonomy, authority, education, dignity, and belonging. These norms no longer function primarily as complete philosophical systems. They survive through families, schools, professions, organizations, rituals, and ordinary judgments about what makes a person capable, respectable, useful, or worthy.

Artificial intelligence may therefore be technologically global while its social meaning remains culturally and institutionally local.

The central problem of the AI age is not only whether machines will replace human labor. It is how societies will preserve human capability, agency, and recognized worth when productive output becomes less reliable as evidence of independent mastery and less secure as a foundation of identity.

Productive Identity as a Modern Presumption

Modern societies formally recognize many sources of human worth. Individuals possess legal rights, moral standing, freedom of conscience, citizenship, family identities, and membership in communities. Yet occupation carries an unusually heavy burden in everyday social life.

The question “What do you do?” often functions as a socially acceptable version of “Who are you?”

An occupational title communicates education, income, social position, competence, discipline, and contribution. A doctor is presumed to possess medical knowledge. An engineer is presumed to understand the system being designed. A writer is presumed to command language and argument. A programmer is presumed to understand the logic represented by the code. A teacher is presumed to possess not only information but also the judgment required to guide another person’s development.

These presumptions are never perfectly accurate. Credentials can exaggerate capability, division of labor can obscure individual contribution, and institutions often reward status independently of merit. Productive identity is therefore not a universal description of how every person experiences work. It is an institutional ideal that modern societies realize unevenly.

Even in partial form, however, the ideal has allowed productive institutions to perform several social functions at once. Work allocates income, but it also allocates recognition. Education develops knowledge, but it also certifies readiness for social participation. Occupation organizes production, but it also supplies routine, community, status, and a narrative of personal development.

The modern individual is therefore not only an autonomous person. The modern individual is expected to become a capable and productive person.^[2]

This expectation represented an important historical achievement. Productive identity allowed individuals to seek status through education, work, and acquired skill rather than relying entirely on inherited rank, family position, or religious hierarchy. It supported mobility, independence, and the belief that people could improve their circumstances through capability and effort.

But it also created a vulnerability. When social recognition becomes closely attached to productive contribution, any technology that weakens the relationship between individuals and their output threatens more than employment. It destabilizes one of the central mechanisms through which modern society interprets and recognizes the person.

AI as a Signaling Shock

Artificial intelligence is not the first technology to externalize human capability.

Writing externalized memory. Printing externalized reproduction. Libraries externalized the preservation of knowledge. Calculators externalized arithmetic. Search engines externalized retrieval. Modern organizations externalized coordination into procedures, databases, and specialized roles.

Human intelligence has always been distributed across people, tools, symbols, and institutions.^[3] There was never a completely self-contained individual standing apart from technological assistance.

AI nevertheless introduces a distinctive change. Earlier tools generally left the boundary between human command and technological execution relatively visible. A calculator produced an answer, but it did not ordinarily produce an entire mathematical explanation in the voice of the user. A search engine retrieved documents, but it did not usually synthesize them into a finished argument that could be presented as the user’s own. A word processor improved the production of text, but it did not generate the structure, reasoning, and language of the document.

Generative AI operates much closer to the visible surface of competence. It produces artifacts that have historically served as signals of the producer’s underlying capability.

An essay once provided evidence of reading, reasoning, organization, and writing. A computer program provided evidence of technical understanding. A legal memorandum provided evidence of research, interpretation, and professional judgment. A visual design provided evidence of imagination, technical execution, and aesthetic choice.

AI does not abolish signaling. It changes its cost and reliability.

A person with limited underlying mastery can now produce a polished artifact at much lower cost. The old artifact therefore becomes less strongly correlated with the capability it previously represented. In economic terms, AI devalues some established signals by making them easier to produce without the corresponding level of independent command.

This does not mean that no capability is involved. Human contribution exists on a continuum. One person may define the problem, supply context, select among alternatives, correct errors, revise language, and assume responsibility for the result. Another may accept the first plausible output with little understanding. Both may present artifacts of similar visible quality.

Prompting, orchestration, revision, and error correction may themselves become valuable capabilities. But they are not necessarily the same capabilities that the finished artifact previously signaled.

Output quality therefore no longer certifies independent capability as reliably as before. Immediate productivity no longer guarantees long-run formation. Occupational participation no longer implies full command of the occupation.

AI is not only a productivity shock. It is also a shock to the signal-extraction process through which institutions infer capability from visible performance.

Education, hiring, credentialing, promotion, professional licensing, and authorship must therefore develop new signals. These may include live explanation, oral defense, process documentation, supervised performance, revision history, error diagnosis, transfer to unfamiliar problems, and demonstrated responsibility for the final result.

The challenge is not to determine whether AI was used. In many settings, AI use will become routine and technically difficult to detect. The deeper question is what a person can explain, evaluate, transfer, and defend after the artifact has been produced.

Capability Extension and Capability Stranding

The weakening of old signals does not mean that AI necessarily diminishes human capability.

Some uses clearly extend it. A researcher can explore unfamiliar literature more quickly. A programmer can understand a new codebase. A student can receive immediate explanations adapted to a specific misunderstanding. A professional can test alternatives that would previously have required far more time or resources.

Experimental evidence already shows that generative AI can raise immediate productivity and, in some settings, produce especially large gains for less experienced workers.^[4] This is important. AI can lower barriers to competent performance and distribute forms of assistance that were previously scarce or expensive.

Other uses improve the immediate artifact while weakening the person or institution behind it. A student submits polished work but cannot explain the argument. A junior professional produces an acceptable report without learning the process through which professional judgment develops. An organization automates routine analytical work and gradually eliminates the entry-level tasks through which future experts were previously trained.

The relevant distinction is therefore not between AI use and nonuse. It is between the directions in which AI use changes underlying capability.

Capability extension occurs when AI enlarges what a person can understand, evaluate, transfer, and ultimately command.

Capability stranding occurs when AI raises current performance while weakening the development, retention, independent exercise, or reproduction of the underlying capability.

These are not mutually exclusive categories. The same AI use may extend one capability while stranding another. A user may gain access to more advanced conceptual analysis while losing fluency in routine execution. A professional may improve error detection while becoming less capable of producing a first draft independently. A student may understand a difficult idea during an assisted session without retaining the ability to reconstruct it later.

The distinction must therefore be evaluated across several dimensions.

First is comprehension. Can the person explain the result and the reasoning behind it?

Second is error detection. Can the person identify when the system is wrong, incomplete, or inappropriate?

Third is transfer. Can the underlying principle be applied to a new problem rather than repeated only in the original setting?

Fourth is retention. Does capability persist after the immediate interaction with the system ends?

Fifth is reproduction. Can the knowledge still be taught, transmitted, and developed across cohorts?

Sixth is independent recovery. Can the individual or organization function when the system fails, changes, or becomes unavailable?

A use of AI may strengthen some of these dimensions while weakening others. The key question is whether repeated use raises current output while allowing the underlying capacity for independent judgment and future formation to decay.

Capability stranding can remain hidden because present performance may improve even as future resilience deteriorates. An organization may appear more productive while becoming less capable of operating outside a narrow technological environment. A profession may appear more efficient while weakening its training pipeline. A school may produce better assignments while developing weaker independent thinkers. A society may enjoy an abundance of competent-looking artifacts while losing the human carriers capable of judging their quality.

This is why the effects of AI cannot be evaluated through productivity statistics alone.

The long-run issue is not only how much output a system generates. It is what kind of human capability the system continues to form, retain, and reproduce.

The Crisis Beneath Employment

Public debate usually begins with employment displacement because jobs are measurable and politically visible. But employment is only the outer layer of productive identity.

Work provides income, but it also supplies routine, status, recognition, social connection, and a sense of contribution. Occupations organize the stories people tell about themselves. They provide a vocabulary for explaining one’s place within society and a structure through which effort accumulates into expertise.

These functions are distributed unevenly. Some workers receive status and autonomy, while others experience work mainly as necessity, insecurity, or subordination. Yet even an unequal occupational system provides socially legible roles through which contribution and competence can be recognized.

AI may weaken those roles even when employment continues.

A professional may retain the title but lose control over the central activity associated with it. The visible occupation survives while the human role changes from producer to supervisor, from drafter to editor, from analyst to validator, or from decision-maker to operator of an intelligent system.

This is not entirely new. Physicians have already lost autonomy to administrative systems, professors to standardized procedures, and professionals to organizational division of labor. AI may accelerate a longer process in which occupational titles persist while independent control over core activities declines.

What does it mean to be a writer when language can be generated on demand?

What does it mean to be a programmer when large portions of code are produced by a model?

What does it mean to be an analyst when synthesis and interpretation are increasingly automated?

What does it mean to be a teacher when explanation can be personalized and delivered instantly by a machine?

These occupations will not necessarily disappear. But their relationship to capability, authorship, responsibility, and personal identity will change.

The deeper question is therefore not only what work will remain for humans. It is what will count as meaningful human capability when high-quality output can be generated without full independent command.

Beneath that lies a more difficult problem: on what basis will societies assign dignity and belonging when scarce human capability no longer anchors social recognition as securely as before?

Formal Modernity and Living Cultural Sediment

Modern societies are formally organized through constitutions, laws, markets, contracts, corporations, bureaucracies, and secular systems of education. These institutions appear to represent a decisive break from older religious, philosophical, and civilizational orders.

Yet formal change does not erase the cultural foundations beneath social behavior.

Societies continue to carry inherited understandings of family obligation, legitimate authority, moral responsibility, education, sacrifice, vocation, harmony, autonomy, and respectable conduct. These understandings shape behavior even when people no longer consciously subscribe to the philosophical systems from which they emerged.

Several concepts should be distinguished.

Culture is the broad field of symbols, beliefs, practices, and meanings through which a society understands itself.

Cultural sediment is the historically inherited stock of norms that persists after the philosophical or religious system from which it emerged has lost intellectual dominance.

Informal institutions are the present rules, expectations, sanctions, conventions, and practices through which those inherited norms affect behavior.^[5]

Carriers are the families, schools, professions, organizations, communities, and rituals that reproduce, reinterpret, or transform those norms across time.

Cultural sediment is therefore not passive. It does not remain influential merely because it was deposited in the past. It survives only when carriers continue to transmit it, often in revised form.

The sediment appears in ordinary expectations rather than formal doctrine. It appears in how parents interpret educational success, how employees respond to hierarchy, how communities judge sacrifice, how leaders perform legitimacy, how families distribute obligation, and how societies distinguish respectable dependence from unacceptable failure.

A modern political inauguration may derive its legal authority entirely from constitutional procedures while continuing to employ sacred texts, inherited oaths, and religious symbolism. The sacred text does not formally transfer power. It supplies a moral language through which the transfer is made socially meaningful.

The same layering appears across formally modern societies. Contemporary schools may operate through standardized curricula while depending on older cultural beliefs about education and authority. Corporations may adopt modern management structures while reproducing inherited expectations concerning hierarchy, loyalty, and relational obligation. Legal systems may recognize individual rights while everyday behavior remains strongly shaped by family responsibility or community judgment.

Modernity rewrote many formal rules. It did not erase the cultural systems through which those rules are interpreted and lived.

Historical Traditions of the Embedded Human

At a high level of abstraction, medieval Western and major Eastern traditions developed different versions of the embedded human being.

The comparison must remain structural rather than doctrinal. Medieval Christian thought, Confucianism, Daoism, Buddhism, and other traditions differ fundamentally in their accounts of reality, personhood, morality, authority, and the ultimate purpose of life. They should not be treated as interchangeable.

Yet many of these traditions located the individual within an order that preceded personal choice.

Medieval Western thought commonly situated the person within a divinely ordered moral universe. Human beings possessed dignity and responsibility, but they did not invent the moral structure of existence independently.

Major East Asian traditions placed differing forms of emphasis on relational identity, family obligation, ritual, cultivation, harmony, and inherited social roles. The individual was understood not only as an isolated chooser but also as a participant in relationships extending across families, communities, and generations.

The common feature is not theology, metaphysics, hierarchy, or political organization. It is the refusal to define the person solely as an autonomous producer whose worth is established through individual output.

Meaning could arise through relationship, obligation, moral cultivation, service, ritual, community, and participation in a larger order.^[6]

These traditions should not be romanticized. Embeddedness can supply meaning, but it can also enforce conformity, exclusion, hierarchy, and inherited roles from which individuals cannot easily escape.

Their relevance to AI is not that modern society should restore them as complete systems. Their relevance is that they preserve historically durable conceptions of human worth that are not grounded entirely in market production, occupational success, or individual cognitive superiority.

Three Forms of Human Worth

The alternatives to productive identity are not exclusively premodern.

Modernity itself created powerful foundations of human worth through universal rights, democratic citizenship, equality before the law, freedom of conscience, expressive individuality, personal autonomy, and civic participation.

The challenge is that human worth operates at more than one level.

Formal worth consists of legal personhood, rights, equality, and citizenship. It establishes that moral and political standing should not depend on productivity.

Social worth consists of recognition, membership, status, trust, and a respected role within a community.

Experiential worth consists of competence, contribution, agency, purpose, and the lived sense that one’s actions matter.

Modern societies have built their strongest protections around formal worth. A person does not lose legal standing because another person is more productive. Human dignity, at least formally, is not earned through market contribution.

But formal dignity does not automatically produce social or experiential worth.

Legal personhood does not by itself supply a respected role. Citizenship does not necessarily provide a daily experience of contribution. Abstract equality may be insufficient when individuals lose the institutions through which they experience competence, purpose, membership, and recognition.

This is why productive identity has carried so much social weight. Occupation has often translated abstract personhood into visible contribution and lived recognition.

If AI weakens occupational identity, formal dignity remains necessary but may not be sufficient. Societies will need other institutions capable of turning personhood into social membership and experienced agency.

The reconstruction of human worth in the AI age will therefore draw from several historical layers at once. It may combine modern personhood and democratic membership with family obligation, religious dignity, community participation, care, service, creativity, moral cultivation, and intergenerational responsibility.

A value that exists only as a philosophical proposition may not be enough. It must be translated into recognized roles, practices, and relationships.

Historical Warning and Institutional Choice

Marx correctly recognized that industrial machinery could absorb human skill into capital, separate workers from control over production, and produce severe social dislocation even while increasing aggregate productivity. He was less successful in predicting the final trajectory of industrial society. Capitalism proved more adaptive, while political reform, mass education, labor organization, regulation, social protection, and the creation of new occupations prevented some diagnosed tendencies from becoming the final social equilibrium. The lesson for AI is neither to dismiss contemporary alarmists nor to accept their forecasts as destiny. Warnings can shape institutional responses, and institutional responses can prevent warnings from becoming prophecies.^[7]

AI itself does not determine whether capability will be extended or stranded. Model developers decide what systems optimize. Firms decide which tasks to automate, which workers to retain, how performance is measured, and who captures the gains. Schools decide whether AI replaces practice or supports it. Professional organizations decide which forms of responsibility and apprenticeship remain necessary. Governments decide which uses require transparency, accountability, or restriction.

The balance between capability extension and capability stranding is therefore not an inherent property of the technology. It is produced through institutional and political choices.

Those choices will not be made from equal positions. Some workers, organizations, and societies will possess greater power to determine how AI is deployed, while others will be required to adapt to systems designed elsewhere. The preservation of agency will therefore be shaped not only by culture but also by bargaining power, ownership, access, and institutional capacity.

These distributional questions deserve separate treatment. Their relevance here is direct: the ability to maintain productive identity depends partly on whether people retain meaningful control over the systems that reorganize their work.

Different Cultural Substrates

AI is entering formally modern societies that do not share identical informal institutions.

This does not mean that civilizations will respond as unified blocs. Every society contains regional, class, generational, religious, and political differences. Cultural traditions overlap, evolve, and compete. Global education, markets, media, and technology have also produced substantial convergence.

The claim is narrower.

When AI weakens productive identity, individuals and institutions will draw differently upon the sources of dignity, responsibility, and belonging available within their inherited social environments.

In societies shaped strongly by productive individualism, AI may be experienced primarily as a threat to authorship, occupational independence, personal achievement, expertise, and control over one’s work.

In settings where relational traditions remain stronger, the same disruption may be interpreted more through family advancement, educational status, role fulfillment, intergenerational mobility, or collective stability.

Where religious sediment remains influential, debate may place greater weight on human uniqueness, vocation, moral responsibility, sacred personhood, or limits on machine authority.

These orientations may also produce different interpretations of the same practice.

Where an essay is treated mainly as evidence of individual authorship and achievement, undisclosed AI assistance may be judged primarily as a misrepresentation of personal capability.

Where education is understood more strongly as preparation for a social or professional role, the central question may be whether the learner has acquired the judgment and responsibility required by that role, even if AI participated in producing the artifact.

The distinction should not be exaggerated. Both concerns can coexist in any educational system. But they direct attention toward different failures. One emphasizes authenticity of individual production. The other emphasizes adequacy of formation for future responsibility.

These are hypotheses about moral vocabularies and informal expectations, not predictions that every member of a society will behave alike.

A technological shock does not arrive in an empty social space. People interpret it through inherited ideas about what a person owes, what a person deserves, how capability should be demonstrated, and what makes a life respectable.

From Cultural Claim to Observable Evidence

Cultural sediment is difficult to identify cleanly because formal and informal institutions evolve together.

Countries with different cultural histories also differ in labor law, educational systems, political institutions, economic structure, language, industrial policy, and access to technology. A comparison between two nations cannot attribute different reactions to cultural inheritance alone.

The strongest evidence may therefore come from narrower comparisons.

The same profession may adopt different AI disclosure norms across countries while retaining similar technical standards. A multinational firm may introduce the same AI system across offices with different organizational cultures. Immigrant communities may respond differently while operating under a common legal system. Different generations within the same society may interpret identical AI practices through different inherited expectations.

These comparisons must also hold bargaining position and control over AI deployment roughly constant, since groups that differ in inherited norms may also differ in their power to determine how the technology is used.

These comparisons will not eliminate the identification problem, but they can make the cultural mechanism more visible.

The argument would be weakened if groups with substantially different inherited norms but broadly comparable immediate institutions responded almost identically to AI-related disruptions. It would gain support if differences became most visible where formal rules left room for interpretation, discretion, stigma, legitimacy, or informal enforcement.

Several questions can guide investigation.

How do educational institutions verify mastery when polished output no longer signals independent capability?

Which professions preserve apprenticeship when routine execution is automated?

What disclosure and authorship norms emerge where formal rules remain silent?

Which communities can confer dignity and status outside conventional employment?

These questions do not classify entire civilizations. They examine how inherited norms influence the interpretation and absorption of a shared technological shock.

AI may therefore reveal more than the future of work. By weakening one of modern society’s dominant sources of identity, it may expose the deeper moral and cultural institutions that have continued to operate underneath formal modernity.

The Institutional Agenda

The argument points toward an institutional agenda rather than a complete policy program.

Education must develop assessments that distinguish polished output from actual mastery while using AI to extend rather than replace learning. Demonstrated explanation, transfer, revision, and error diagnosis may become more important than possession of a finished artifact.

Professional organizations must preserve judgment, responsibility, and apprenticeship when execution becomes increasingly automated. The disappearance of routine entry-level work should not be allowed to eliminate the formation path through which future experts emerge.

Firms must decide whether AI is being used to develop employee capability or only to extract more output while allowing internal knowledge to decay. Short-run productivity should not be confused with long-run organizational competence.

Governments and standard-setting bodies must determine where transparency, accountability, and human responsibility remain necessary. AI should not be treated as an external force to which institutions can only react. Regulation, public investment, procurement, professional standards, and education policy can shape how it develops and where it is deployed.

Communities must sustain recognizable forms of contribution and belonging that are not measured solely through market productivity. Care, mentorship, civic participation, cultural transmission, and intergenerational support may need stronger forms of public recognition.

These are separate questions requiring fuller treatment. Their common objective is to prevent AI-assisted abundance from being purchased through the erosion of human agency, capability, and social membership.

Conclusion

Artificial intelligence is exposing a weakness in the modern organization of identity.

Productive output has been asked to carry too much: income, status, competence, independence, dignity, and personal meaning. As AI lowers the cost of generating sophisticated artifacts and separates visible performance from independent capability, this arrangement becomes less stable.

That does not mean that work will disappear, that human capability will inevitably erode, or that the darkest predictions of technological displacement will come true. AI can extend capability as well as strand it. It can broaden access as well as concentrate control. It can create new forms of participation as well as weaken established ones.

The outcome will depend on choices made by schools, firms, professions, governments, communities, model developers, and users. These choices will determine which tasks are delegated, which capabilities continue to be formed, which signals replace the old ones, who retains responsibility, and who captures the gains.

The decisive conflict will not be between human beings and intelligent machines in the abstract. It will be between institutional arrangements that treat people mainly as sources of output and those that continue to cultivate capability, responsibility, agency, and recognized contribution.

Those choices will not be made on a culturally empty field.

Societies will draw upon cultural sediments accumulated over centuries: religious conceptions of dignity, relational traditions of obligation, modern ideals of autonomy, civic understandings of membership, and inherited expectations about education, authority, authorship, and contribution.

These inheritances will not determine the future mechanically. They will shape which losses societies notice, which forms of dependence they tolerate, which institutional responses they regard as legitimate, and which sources of human worth they are prepared to recognize.

The central challenge of the AI age is therefore not only to preserve employment or maximize productivity. It is to construct a social order in which people can continue to develop capability, exercise agency, and possess formal, social, and experiential worth even when intelligent machines perform an increasing share of productive execution.

AI may be technologically universal. The reconstruction of human worth will remain culturally and institutionally local.

Endnotes

[1] The interpretation of visible output as a signal of an underlying but unobservable capability follows the classic signaling framework developed by A. Michael Spence. See Spence, “Job Market Signaling,” Quarterly Journal of Economics 87, no. 3 (1973): 355–374. The AI-era problem differs from Spence’s original model because the technology changes the cost and reliability of producing the signal itself. Back

[2] On the historical construction of modern identity around autonomy, vocation, work, and social recognition, see Charles Taylor, Sources of the Self: The Making of the Modern Identity (Cambridge, MA: Harvard University Press, 1989); and Robert N. Bellah et al., Habits of the Heart: Individualism and Commitment in American Life (Berkeley: University of California Press, 1985). Back

[3] The idea that cognition depends upon external objects, social systems, and material environments is developed in Andy Clark and David J. Chalmers, “The Extended Mind,” Analysis 58, no. 1 (1998): 7–19; and Edwin Hutchins, Cognition in the Wild (Cambridge, MA: MIT Press, 1995). These approaches support the claim that cognitive externalization is not unique to AI while leaving open whether different forms of externalization strengthen or weaken independent human command. Back

[4] Evidence that generative AI can increase immediate productivity and, in some settings, disproportionately assist less experienced workers appears in Shakked Noy and Whitney Zhang, “Experimental Evidence on the Productivity Effects of Generative Artificial Intelligence,” Science 381, no. 6654 (2023): 187–192; and Erik Brynjolfsson, Danielle Li, and Lindsey R. Raymond, “Generative AI at Work,” Quarterly Journal of Economics 140, no. 2 (2025): 889–942. These findings establish the possibility of capability extension but do not resolve the longer-run questions of retention, independent command, and skill reproduction. For the broader automation context and the distinction between technologies that displace existing tasks and those that create or reinstate human tasks, see David H. Autor, “Why Are There Still So Many Jobs? The History and Future of Workplace Automation,” Journal of Economic Perspectives 29, no. 3 (2015): 3–30; and Daron Acemoglu and Pascual Restrepo, “Automation and New Tasks: How Technology Displaces and Reinstates Labor,” Journal of Economic Perspectives 33, no. 2 (2019): 3–30. Back

[5] The distinction between formal rules and informal constraints follows Douglass C. North, Institutions, Institutional Change and Economic Performance (Cambridge: Cambridge University Press, 1990); and North, “Institutions,” Journal of Economic Perspectives 5, no. 1 (1991): 97–112. “Cultural sediment” is used here for historically inherited norms that continue to operate through present carriers, sanctions, expectations, and practices after their original philosophical or religious systems have lost intellectual dominance. Back

[6] The comparison between older traditions concerns the structural idea of an embedded person rather than doctrinal equivalence. For Western accounts of moral identity situated within inherited practices and traditions, see Taylor, Sources of the Self; and Alasdair MacIntyre, After Virtue: A Study in Moral Theory, 3rd ed. (Notre Dame, IN: University of Notre Dame Press, 2007). For a Confucian account of relational and cultivated selfhood, see Tu Weiming, Confucian Thought: Selfhood as Creative Transformation (Albany: State University of New York Press, 1985). These traditions differ substantially in metaphysics, theology, and political implications. Back

[7] Marx’s discussion of machinery absorbing accumulated social knowledge and shifting the worker toward supervision appears most directly in the “Fragment on Machines” in Karl Marx, Grundrisse: Foundations of the Critique of Political Economy, trans. Martin Nicolaus (London: Penguin, 1973), especially 690–712. His broader treatment of machinery and the labor process appears in Capital, vol. 1, chap. 15. The subsequent history of industrialization illustrates why a diagnosed technological tendency should not be treated as a complete prediction of the eventual social equilibrium. Back