By Faith Cheltenham
There is something breathtakingly elitist buried inside an otherwise interesting essay about artificial intelligence and the future of work. In “The Eternal Complement,” Hemanth Asirvatham and Elliott Mokski ask us to imagine superintelligence as a civilization containing a billion Einsteins, then write that such a civilization “still needs most of them to mine the quarries and manage the accounting.” Their larger argument is more nuanced than that sentence alone: they argue that civilization depends upon enormous systems of execution, institutions and physical labor, and they ultimately imagine machine intelligence taking over much of the monotonous work required to turn ideas into reality. They even acknowledge that AI could allow people previously excluded by institutional gatekeepers to pursue ideas themselves (Asirvatham & Mokski, 2026). But their billion-Einsteins metaphor still stopped me cold because it carries an assumption so old that we barely notice it anymore: apparently even in a civilization overflowing with genius, most beings must still be assigned to support the ambitions of a comparatively small number of beings doing the interesting stuff.
Why exactly should Einstein be mining the quarry? More importantly, why have we constructed a civilization in which somebody may have to spend most of a human lifetime breaking rock so somebody else has enough freedom to contemplate the universe? The authors are correct that no great human accomplishment is really individual. An astronomer depends upon telescope makers, machinists, teachers, nurses, farmers, electricians, accountants, truck drivers and thousands of people she will never meet. Where I part company with this vision is in treating those people principally as complements to somebody else’s achievement. The person welding the telescope screw may also have a theory about the stars. The nurse may be a novelist. The quarry worker may have the mathematical intuition of the astronomer but never have encountered the combination of time, education, security and institutional permission necessary to discover it. Research on collective intelligence already complicates the idea that concentrating supposedly exceptional individuals necessarily creates the strongest problem-solving system; work by Hong and Page found conditions under which diverse groups of problem solvers can outperform groups selected solely for individual ability, while Woolley and colleagues found evidence that group performance reflects a collective-intelligence factor not reducible to the intelligence of the group’s smartest member (Hong & Page, 2004; Woolley et al., 2010).
This is where my Suresha–Cheltenham Model and the larger framework I call Cultural Physics become useful beyond their original application to institutional inclusion. Cultural Physics asks us to stop looking only at who is physically present inside a system and measure how representation, power, resources, mobility, trust and bias actually move through it. I developed the model from Ron Suresha’s earlier work turning bisexual representation claims into auditable criteria, expanding that idea into a mathematical framework concerned not simply with representation but with who decides, who receives resources, who advances, who is trusted and where power accumulates (Suresha, 2013; Cheltenham, 2026). Power in organizations has long been understood as relational rather than merely personal, and network position and dependency affect who actually exercises influence (Emerson, 1962; Brass, 1984). Cultural Physics takes that insight seriously enough to ask what happens when we apply it to the entire political economy.
Consider the billion Einsteins as a mathematical exercise rather than a metaphor. Suppose every person begins with identical intellectual potential, I = 1, giving us one billion units of potential intelligence. Now imagine that 900 million of those people spend the overwhelming majority of their available time securing food, housing, healthcare, transportation and survival while 100 million possess enough resources to devote substantial time to experimentation, scholarship and invention. The civilization still contains one billion Einsteins biologically, but it does not have one billion Einsteins functionally. Its realized intellectual capacity is constrained by access. If the first group can devote only 10 percent of its potential intellectual time to unconstrained creation while the second can devote 80 percent, the crude thought experiment gives us (900 million × .10) + (100 million × .80) = 170 million Einstein-equivalents of usable creative time. Nothing happened to anybody’s intelligence. The system simply stranded 830 million potential units through allocation.
Change the allocation and the civilization changes without making a single person smarter. If automation and broad material security allowed everybody to devote even 50 percent of their available capacity to chosen intellectual, artistic, scientific, civic or entrepreneurial activity, the same population would yield 500 million Einstein-equivalents under this intentionally simplified exercise. That is nearly three times the realized capacity of the unequal scenario. These numbers are not empirical measurements or predictions; they are a thought experiment exposing a variable our stories about genius routinely hide. We obsess over the quantity of intelligence while ignoring the coefficient determining whether intelligence gets to move.
The equations I have been developing for Cultural Physics make the same point another way. One experimental expression models Inclusion Motion as R × P × M × T × A − D, where representation, power, mobility/resources, trust/belonging and adjacency/proxy legitimacy interact while bias creates drag (Cheltenham, 2026). Multiplication matters. If one essential dimension approaches zero, abundance elsewhere cannot simply erase its absence. A society might therefore have astonishing technological representation—AI everywhere—and extraordinary aggregate wealth while ordinary people possess little mobility or decision power. Calling that civilization abundant because its total productive output is enormous would make the same conceptual mistake as calling an institution inclusive because its workforce photograph is diverse while all meaningful decisions remain concentrated upstairs.
That is also why my model includes a non-substitution rule: when a claimed population has neither direct representation nor a validated proxy, an institution’s overall inclusion score is capped. I am extending that logic here from populations to dimensions. If representation is excellent but power is effectively zero, something important remains broken. If technological productivity becomes extraordinary while economic mobility collapses, the machines have not solved the human problem. If artificial intelligence generates trillions of dollars of value but ownership of that productive machinery becomes extraordinarily concentrated, aggregate abundance can coexist perfectly well with individual dependence. Cultural Physics describes concentrated power almost gravitationally: existing centers of power pull decisions, careers, resources and legitimacy toward themselves, while institutional inertia makes those arrangements difficult to dislodge (Cheltenham, 2026). This is metaphor, not literal Newtonian physics, but it forces us to identify the variables instead of waving toward “the economy” as though it were weather.
Once we think this way, the quarry problem looks completely different. The obvious answer to a civilization containing a billion Einsteins that still needs stone is not to decide which 900 million Einsteins deserve quarry duty. Automate the damn quarry. Machines should perform dangerous, repetitive and physically destructive work wherever that can be accomplished safely. Some human beings will still choose construction, welding, farming, logistics, accounting and thousands of occupations routinely described as ordinary because they enjoy them, value them or find meaning in mastering them. There is enormous dignity in those forms of work. The indignity enters when survival requires a person to surrender so much time and possibility that the rest of their capacities never get a chance to exist.
This leads to the class strangely missing from so many AI thought experiments: the extremely rich. We endlessly ask what happens to workers when artificial intelligence eliminates labor, yet rarely reverse the question and ask what happens to billionaires when artificial intelligence eliminates scarcity. If intelligence becomes abundant, automation becomes abundant and productive capacity increases enormously, what economic principle requires preserving a system in which a tiny number of human beings possess extraordinary claims over the productive machinery everybody else increasingly depends upon? Research and policy analysis of AI already recognizes that ownership matters: IMF staff analysis has warned that AI’s distributional consequences depend partly on how gains to capital interact with labor income, potentially increasing wealth inequality when capital returns accrue disproportionately to high earners (Cazzaniga et al., 2024). The technological question therefore cannot be separated from the ownership question.
So yes, perhaps we should get rid of rich people instead.
I mean that economically, not physically. I am talking about eliminating extreme wealth as a social class capable of accumulating quasi-governmental power, not eliminating human beings who currently possess wealth. People can build companies, become successful, own property, invent wonderful things, leave meaningful inheritances to their children and buy ridiculous shoes because ridiculous shoes make them happy. Freedom does not require universal poverty or forced sameness. But neither does freedom require allowing individual fortunes to become so enormous that their owners can command quantities of labor, housing, media, infrastructure and political access comparable to institutions. The relevant question is not whether somebody is permitted to become prosperous. It is how much gravitational mass we allow wealth to acquire before everyone else’s choices begin bending around it.
The great promise of artificial intelligence should therefore be larger than determining which human tasks machines cannot perform. Humanity may not suffer from a shortage of intelligence at all. We may suffer from a catastrophic shortage of permission to use intelligence. Education is rationed by price and geography; laboratories and computing power have historically been concentrated inside institutions; housing insecurity consumes cognitive and temporal resources; caregiving takes time; illness takes time; transportation takes time; bureaucracies take time; poverty takes enormous amounts of time. Once we construct an economy requiring billions of human beings to exchange most of their waking hours for survival, it becomes almost absurd to look around afterward and wonder why there aren’t more Einsteins. Perhaps there are. They’re working.
Artificial intelligence gives us an extraordinary opportunity to change the coefficients rather than merely increase the numerator. A Black girl in South Los Angeles with an engineering idea should not need twenty institutional permissions before she can test it. A disabled programmer should not need venture capitalists to determine whether an invention deserves to exist. A mother should not have to choose between feeding her children and spending six months investigating the idea living inside her head. A farmer in Ghana should be able to combine local knowledge with computational capacity without surrendering ownership of the resulting invention to a multinational corporation. AI could become one of history’s greatest decentralization technologies because dramatically lowering the cost of expertise, computation and coordination can allow individuals and small groups to attempt work once reserved for wealthy organizations. Even “The Eternal Complement” recognizes this possibility when its authors describe AI reducing the organizational infrastructure required for ambitious people to pursue ideas (Asirvatham & Mokski, 2026).
Progress really does call on everyone. Where I disagree is with imagining everyone as a component in a humming economic machine whose ultimate purpose is enabling somebody else’s frontier innovation. Cultural Physics asks a different question: who actually possesses enough power, resources, mobility and trust to move? The welder is not merely the complement to the astronomer because the welder is also a complete human being. The accountant does not exist to make genius administratively possible. The nurse who treated the astronomer possesses desires, curiosities, memories, talents and unrealized possibilities entirely unrelated to the economic service the astronomer happened to receive from her.
The most interesting question of the Intelligence Age may therefore have very little to do with which occupations survive artificial intelligence. For thousands of years, scarcity meant somebody had to carry the stone. Increasingly, machines can carry it. A civilization wealthy enough to build superintelligence should finally be wealthy enough to ask what human beings might become when being economically necessary is no longer the admission price for being allowed to live.
Give me the billion Einsteins, then. Give me a billion artists, mechanics, mothers, mathematicians, gardeners, programmers, oddballs, welders, poets and people who spend Tuesday repairing an engine and Wednesday wondering about the origin of gravity. I don’t want a civilization that discovers a billion Einsteins and immediately sends most of them underground to fetch rocks for the lucky Einsteins upstairs.
That isn’t an intelligence problem.
It’s an allocation problem.
And perhaps the smartest thing superintelligence could finally teach us is that we have been wasting human intelligence all along.
References
Asirvatham, H., & Mokski, E. (2026). The Eternal Complement. OpenAI.
Brass, D. J. (1984). Being in the right place: A structural analysis of individual influence in an organization. Administrative Science Quarterly, 29(4), 518–539.
Cazzaniga, M., Jaumotte, F., Li, L., Melina, G., Panton, A. J., Pizzinelli, C., Rockall, E. J., & Tavares, M. M. (2024). Gen-AI: Artificial Intelligence and the Future of Work. International Monetary Fund.
Cheltenham, F. (2026). The Suresha–Cheltenham Model for Inclusion and Cultural Physics. Working manuscript.
Emerson, R. M. (1962). Power-dependence relations. American Sociological Review, 27(1), 31–41.
Hong, L., & Page, S. E. (2004). Groups of diverse problem solvers can outperform groups of high-ability problem solvers. Proceedings of the National Academy of Sciences, 101(46), 16385–16389.
Suresha, R. J. (2013). Where’s the “B” in national LGBTQ organizations? LGBTQ Nation.
Woolley, A. W., Chabris, C. F., Pentland, A., Hashmi, N., & Malone, T. W. (2010). Evidence for a collective intelligence factor in the performance of human groups. Science, 330(6004), 686–688.
