AI disclosure

The Human Prosperity AI Manifesto

A framework for using artificial intelligence to expand prosperity without eliminating human economic participation

Humanity should use artificial intelligence to expand human capability, reduce scarcity, improve safety, accelerate discovery, and raise living standards. We should not treat the elimination of human participation as the default measure of technological success.

AI can perform work faster, cheaper, and at enormous scale. That capability is valuable. But an economy is not merely a production system. Human beings are simultaneously workers, consumers, taxpayers, borrowers, homeowners, entrepreneurs, students, parents, investors, and citizens. When labor income is removed at scale, the effects propagate through every one of those roles.

The governing principle of the AI economy should therefore be simple: use AI to eliminate scarcity, waste, dangerous work, repetitive work, preventable error, and unnecessary delay, not to eliminate people from economic life.

We affirm the following principles:

1. Human prosperity is the objective. Productivity is a means, not the final measure of success.
2. Augmentation should be preferred to replacement when both can produce competitive outcomes.
3. Productivity gains should be shared through some combination of higher real income, lower prices, shorter working hours, broader ownership, improved services, and better quality of life.
4. Entry-level work must evolve rather than disappear. Every economy requires pathways through which novices become experts.
5. Workers displaced by technology require real transitions into occupations with durable demand, not generic retraining for its own sake.
6. Ownership matters. If AI increasingly performs productive labor, ordinary households need broader participation in the capital that generates AI-driven wealth.
7. Competition matters. AI infrastructure, models, compute, energy, data, and intellectual property should not become so concentrated that a small number of institutions control the productive intelligence of society.
8. Education must prepare people to work with AI, challenge it, verify it, supervise it, and exercise the human judgment that automated systems cannot legitimately own.
9. Automation should be deployed aggressively where labor is scarce, work is dangerous, services are inaccessible, or human capacity is being wasted on low-value process.
10. Governments should measure the AI transition using more than GDP and unemployment. Median real income, labor-force participation, underemployment, entry-level hiring, labor’s share of income, hours worked, household wealth, business formation, and distribution of productivity gains are equally important.
11. Companies should recognize that employees are also customers. Production at scale requires purchasing power at scale.
12. Global AI policy must consider developing economies whose competitive advantage has historically depended on human labor. AI should widen access to expertise, not close the development ladder.
13. Human accountability must remain attached to consequential decisions. Efficiency cannot replace responsibility.
14. The benefits of AI should be judged over decades, not merely by the next quarter’s labor-cost reduction.

The central economic challenge of advanced AI is not only whether machines can increase productive capacity. AI may help create greater abundance, improve safety, expand access to expertise, accelerate discovery, and reduce burdensome work. The accompanying challenge is how people continue to participate meaningfully in that prosperity if some forms of human labor become less necessary to production.

A durable AI economy should preserve broad purchasing power and economic participation. Depending on how labor markets evolve, that participation may continue primarily through wages or may be supplemented by broader capital ownership, profit sharing, social dividends, transfers, reduced working hours with sustained income, or institutions not yet developed. The appropriate mix is open to debate and should respond to evidence rather than ideology.

We reject the false choice between technological progress and human economic security. A successful civilization should be capable of achieving both.

AI is already changing work and is likely to change it further. The question is whether institutions, businesses, educators, workers, and governments shape that transition so technology serves human flourishing, or allow short-term incentives to dominate decisions with long-term consequences.

AI should make human beings more capable and expand opportunity, not make economic exclusion an acceptable measure of technological success.

Use AI to reduce scarcity, expand capability, and improve human life, not to treat people as costs to be eliminated.

Enhance the Work, Not Replace the Worker

Artificial intelligence should generally be deployed first as a tool for human augmentation rather than reflexive human substitution. When AI can perform part of a person’s job faster, safer, cheaper, or more accurately, the first question should not automatically be “How can we eliminate this position?” It should also be “How can we use this capability to make people more productive, more capable, safer, and more effective?” AI is particularly valuable when it reduces repetitive administration, dangerous work, delays, preventable errors, and barriers to expertise, while allowing people to devote more time to judgment, creativity, relationships, leadership, accountability, problem solving, innovation, and other high-value activities. Productivity gains should therefore be evaluated not only by labor savings, but also by new capabilities, quality, safety, access, resilience, growth, and human outcomes.

A teacher equipped with AI may be able to provide more individualized instruction. A physician may gain more time for patients. An accountant may spend less time preparing information and more time interpreting it. An engineer may evaluate more design alternatives. A scientist may investigate questions faster. These are examples of AI expanding human capacity when implementation is designed around augmentation, quality, and accountability rather than replacement alone.

This approach also preserves the human-capital ecosystem. People continue gaining experience, developing judgment, mentoring others, transferring institutional knowledge, becoming experts, earning income, paying taxes, supporting families, and participating in the economy as consumers.

The governing objective is: Automate appropriate tasks. Augment people. Expand human capability. Use AI to enhance work and human potential, not to make worker replacement the default objective. Use AI to reduce scarcity and expand opportunity.

The purpose of this manifesto is therefore not to resist AI. It is to argue for ambitious AI development and adoption that remains accountable to human flourishing. Progress should be judged by both what the technology can do and what its deployment enables people and societies to become.

The same power creates responsibility. AI can also be used carelessly or deliberately in ways that concentrate power, displace people without viable pathways forward, amplify error, weaken privacy, enable manipulation, deskill institutions, or create dangerous dependencies. These outcomes are neither inevitable nor imaginary. They are governance, design, market, education, and human-choice problems that should be anticipated and managed.

AI is one of the most powerful general-purpose tools developed in the modern era. Used well, it can extend human intelligence, make expertise more accessible, accelerate scientific discovery, improve education and healthcare, assist people with disabilities, strengthen public services, reduce dangerous and monotonous work, help small organizations compete with larger ones, and enable individuals to create and solve problems at scales previously beyond their reach.

The Promise and Responsibility of AI

New Principles: Human Capability and Resilience

AI must not merely preserve today’s experts. It must preserve the process by which society creates tomorrow’s experts.

Do not allow automation of a critical capability to eliminate the human capacity needed to understand, independently verify, repair, recover, and, when necessary, perform that capability.

We should not teach people less merely because AI can perform more tasks. AI should be used to help people learn more effectively, reach deeper understanding, and develop the judgment required to use powerful tools responsibly.

Human capability is a form of critical infrastructure. A resilient AI economy should maintain sufficient human expertise to supervise and challenge automated systems and to sustain essential functions when those systems fail, produce incorrect results, become unavailable, or become economically inaccessible.

The Capability Crisis

A capability crisis can occur when sustained technological substitution contributes to the population possessing a critical skill falling below the level needed to independently perform, supervise, verify, repair, or recover systems on which society depends.

A capability crisis does not require AI to stop working. It can emerge when humans become progressively less capable of determining whether AI is correct. The danger is therefore not merely technological failure, but human inability to recognize, diagnose, or recover from technological failure.

Poorly designed AI adoption could create a self-reinforcing cycle: AI performs more foundational tasks; some employers reduce entry-level hiring; career pathways narrow; fewer people enter or remain in the discipline; apprenticeship opportunities decline; experienced workers retire; the stock of human expertise shrinks; dependence on automated systems increases; and fewer humans remain capable of independently evaluating the technology. This is a risk pathway, not an inevitable outcome, and deliberate education, role design, mentorship, and apprenticeship can counter it.

The Human Expertise Paradox

The more capable AI becomes, the easier it may become to justify eliminating opportunities for humans to practice foundational tasks. Yet many of those tasks have historically helped future experts develop judgment. The goal should be to redesign learning-rich work rather than preserve obsolete tasks for their own sake.

Foundational knowledge and experiential knowledge are separate assets. Formal education can provide principles, theory, and methods. Expertise requires repeated practice, mistakes, debugging, mentoring, increasing responsibility, and exposure to consequences.

Tacit knowledge is knowledge acquired through experience and practice that is difficult to fully document or teach explicitly. It is the accumulated judgment that allows an experienced professional to recognize that something is wrong even when the formal rules, documentation, or AI recommendation suggest otherwise.

Preserve the Expertise Pipeline

The expertise pipeline is: education -> foundational competence -> supervised practice -> independent practice -> increasingly complex responsibility -> accumulated judgment -> expertise.

Automating substantial portions of entry-level work can create an apprenticeship void if organizations do not replace lost learning opportunities with deliberate practice, mentorship, simulation, and progressively responsible work. In that case, an organization may appear more productive in the short term while drawing down a stock of expertise created under an earlier labor model.

Preserving senior experts while failing to maintain credible pathways for developing their successors can become a form of managed depletion.

The Subsidized Automation Problem

The prices customers pay for AI services do not necessarily reveal the providers’ full long-run economics. A sound comparison should consider, where data are available, training and inference costs, compute infrastructure, data centers, electricity, networking, engineering, depreciation, financing, governance, and capacity expansion, while recognizing that unit costs and business models can change rapidly.

Businesses evaluating substitution should compare the expected total cost and value of AI-enabled work with the expected total cost and value of human or human-plus-AI work. A retail subscription price alone is not a sufficient basis for a long-term workforce decision.

If organizations eliminate human capability on the assumption that today’s AI prices, vendors, and access conditions will persist, later price increases, consolidation, outages, or access changes could create labor-capacity lock-in: dependence on external systems for work the organization no longer has enough people capable of performing independently.

Measure a Successful AI Economy

A successful AI economy is one in which technological progress contributes to broad improvements in human prosperity, capability, economic security, opportunity, resilience, and quality of life, while managing displacement fairly and maintaining meaningful pathways for people to participate in economic and civic life.

Major AI productivity gains should be evaluated by asking where the benefits appear: higher or more secure income, lower prices, shorter or more flexible working hours, broader ownership, better products and services, improved public goods, expanded output, new businesses and occupations, greater accessibility, or higher returns to capital. A healthy system can support several of these outcomes at once.

Cost reduction and consumer-price reduction are not the same. Evidence that AI lowers a firm’s operating cost does not establish that consumers receive an equivalent price reduction.

The ultimate measure of an AI economy should not be how much work machines can perform, but whether their capabilities contribute to better human lives, stronger institutions, broader opportunity, and a more resilient and productive society.

AI can be an extraordinary instrument for good. Whether it becomes one at societal scale depends not only on model capability, but on the incentives, institutions, education systems, business decisions, and values that govern how that capability is used.

Some jobs will change, some will disappear, and new jobs and industries will emerge. The responsible response is neither denial nor fatalism. It is active adaptation: redesign work, educate people for deeper capability, create credible transition pathways, preserve critical expertise, broaden access to the gains from productivity, and maintain human accountability where consequences are significant.

We support continued AI innovation, investment, experimentation, and deployment. The objective is not to freeze existing jobs or preserve inefficient processes. It is to ensure that productivity gains are converted into wider capability, opportunity, safety, knowledge, prosperity, and resilience.

A Human-Centered AI Compact

Evidence and Intellectual Grounding

The concerns in this manifesto are hypotheses and policy principles informed by emerging evidence, not claims that every risk described is already occurring at scale. Recent research and institutional work support closer attention to entry-level learning, apprenticeship, and expertise formation. Conceptual papers should be distinguished from empirical studies and official programs.

U.S. Department of Labor (2026): national initiative to integrate AI skills into Registered Apprenticeship programs, reflecting a policy effort to modernize training while preserving structured pathways into skilled work.

Sarala, S. (2026), “The apprenticeship void: AI augmentation and entry-level displacement in knowledge work,” Vilakshan – XIMB Journal of Management, 23(2), 174–185. Conceptual paper; the framework requires empirical validation.

Dogonowski, R. (2026), “The Apprenticeship Externality: AI and the Intertemporal Supply of Expertise.” SSRN working paper. A theoretical model of how automating entry-level generation tasks could reduce the future supply of judgment and verification capability.

McKinsey & Company (2026), “Building expertise in the age of AI: Who trains the next generation?” Reports that entry-level work is changing and argues for deliberate redesign of roles, learning, coaching, and knowledge management. Its cited employer survey also found mixed effects, including employers expecting both increases and decreases in entry-level hiring.

World Economic Forum (2026), “Artificial Intelligence and the Future of Entry-Level Work: A Framework for Safeguarding and Reinventing Early Career Pathways.” Emphasizes job access, job design, talent pipelines, and education-system alignment rather than assuming entry-level work will simply disappear.