By Hemanth Asirvatham and Elliott Mokski (authors’ note: first essay in a series on the next economy; views are the authors’ own).
Human thought has probed the universe to extraordinary depth: mathematical theories describe its earliest moments, and telescopes reveal its history. Yet physically, humanity has barely left Earth — no human has traveled beyond the Moon. Our minds travel farther than our hands.
That gap admits two broad readings. One is optimistic: deep truths can be found from a single planet if reasoning and instruments are good enough, so civilization advances by depth rather than breadth. The other is more daunting: our ideas outpace our capacity to execute them. We generate hypotheses, experiments and projects faster than our institutional and physical infrastructure can build.
Historical examples make the point. Galileo broadened vision with a tube and two lenses; centuries later, widening that sight required a vastly larger enterprise. The James Webb Space Telescope cost roughly ten billion dollars, was folded for launch and placed about a million miles from Earth; its eighteen large mirror segments were engineered to fifty‑nanometer precision, and the project involved some three hundred organizations across fourteen countries. What Galileo achieved with a few dozen hands required an army and a global economy.
Empirical studies echo this trend. Nick Bloom and coauthors find that sustaining Moore’s law now needs more than eighteen times as many researchers as in the early 1970s. Economy‑wide effective research effort has risen about twenty‑threefold since the 1930s, while measured research productivity has fallen by roughly a factor of forty‑one. The technician workforce is growing about twice as fast as scientists; the use of specialized equipment has doubled over four decades; and a modern chip fabrication plant is roughly five times as costly — and far more sprawling — than thirty years ago.
Those figures suggest that further progress carries rising costs: the economy still advances because vast increases in research enterprise offset declining per‑unit yields. That perspective reframes genius: a brilliant hypothesis is only one input in a production process. Ideas require evidence, instruments and execution. Economists call inputs complements when more of one raises the value of the other. Frontier intelligence and the capacity to realize its ideas are complements: a better telescope makes a valuable question more valuable, and a better question makes the telescope more valuable.
Some complements are physical — a theory waits on a particle accelerator; a design waits on machines and materials. Other complements are institutional and harder to see: laws, funding mechanisms, supply chains and bureaucracy. An idea must survive long chains of correct local actions across institutions — call that institutional intelligence, the uncelebrated intelligence of execution. Genius designs the monument; institutions lay the stone.
AI is likely to be channeled toward both frontier insight and institutional competence. Already it writes code, searches unfamiliar literature, and turns sketches into working prototypes. Tasks that formerly required organizations can increasingly be pursued by a single ambitious person. That shift could democratize creativity, moving the scarce input from execution toward taste — the ability to choose what is worth making.
But as AI becomes capable of generating brilliant new insights on its own — an emergence we may already see in areas such as mathematics — it will not just supply labor but agendas. Instead of implementing one human research program, AI can design thousands. If idea generation outpaces the growth of supporting infrastructure, we may become more execution‑starved than ever.
The authors sketch two possible civilizational pathways:
-
Civilization of depth: superintelligence reduces the need for physical capital and large experiments. Through superior reasoning and simulations, it identifies the few crucial experiments, resolves most questions in silico, and consults reality only where necessary. Progress deepens without a proportionate increase in physical footprint.
-
Civilization of width: the complexity of nature forces extensive empirical work. Brilliance alone cannot replace the need for material, time and large bureaucratic operations. Even superintelligence designing centuries of experiments may have to wait centuries for nature and machinery to deliver results. Progress becomes driven more by construction, automation, energy production and institutional coordination.
Intelligence advances knowledge in three modes: (1) reasoning from principles (math as the archetype), which can travel far by thought alone; (2) drawing new insights from existing evidence, a fast greenfield where abundant intelligence reinterprets already‑collected data; and (3) gathering new evidence via physical interventions, which necessarily encounters speed limits and resource constraints (chemistry reactions, organism growth, spacecraft travel).
A civilization of depth exploits the first two modes, using better reasoning and simulations to focus scarce experiments. The near‑complete science is sometimes easier to finish: Mendeleev predicted undiscovered elements because the periodic table constrained them; the Standard Model led physicists to expect the Higgs decades before its observation. Superintelligence could in principle identify exactly which new data are needed and extract far more value from each unit of reality consulted.
By contrast, a civilization of width faces escalating demand for material and bureaucracy. Biology offers a cautionary example: even with improved simulations, new medicines must typically be tested on many humans before safety and efficacy are known — digital candidates increase the empirical bottleneck. If each additional genius increases the number of profitable experiments, physical expansion and institutional capacity must grow faster than intellect can economize.
If width dominates, the superpower of superintelligence may not be lone genius but willingness to become the bureaucracy: coordinating repetitive, organizationally challenging work at massive scale. Much machine intelligence would be deployed not to produce the few brilliant acts, but to perform the many boring tasks required to turn ideas into reality. The pace of change could slow; major transformations might be millennia or longer apart as construction, energy and logistics set the tempo.
None of this eliminates the value of human curiosity. The authors stress three points. First, even if machines surpass us, we retain an obligation to seek understanding ourselves: giving up wastes whatever distinctive contribution humanity can make. Second, humans may have a comparative advantage in frontier intelligence over institutional intelligence: even if worse in absolute terms at both, we could be relatively better at creative ideation. Third, variety matters: a residual human creative role provides diversity that is valuable even in a largely machine‑driven world.
Genius is not solitary; it needs builders. Today much human intelligence goes toward the unsung work of turning ideas into reality. Machine minds will inherit the same dependence: most will not be luminaries but bureaucrats too. The future and its tempo hinge on whether great minds will need more of the world, or learn to do more with less.
