spb/spboucher.ai Public

1# What Happens When the Cost of Intelligence Approaches Zero?23For most of economic history, intelligence has been expensive.45Not intelligence in the abstract sense, but economically useful cognitive work: analyzing information, writing software, designing products, preparing contracts, conducting research, managing organizations, or making decisions.67These activities required educated humans.89Educated humans are scarce. They require years of training, have limited working hours, and cannot be replicated instantly.1011Artificial intelligence introduces a strange possibility into this system:1213**What happens if useful intelligence stops being scarce?**1415We are still far from intelligence being literally free. Models require chips, electricity, infrastructure, data, and engineering.1617But the marginal cost of many cognitive tasks is already collapsing.1819And economics tells us that when the price of an important input collapses, the consequences are rarely limited to that input.2021The entire system reorganizes around it.2223## Intelligence as an Economic Input2425Consider intelligence as simply another factor of production.2627A simplified economy might contain:2829* capital,30* labor,31* energy,32* natural resources,33* and intelligence.3435Historically, intelligence and labor have been tightly coupled.3637If a company wanted twice as much accounting work, software development, legal analysis, financial modeling, or research, it generally needed more people.3839AI begins to separate these two variables.4041A machine can now produce certain forms of cognitive output without requiring an additional human worker for every additional unit of output.4243This distinction is fundamental.4445Software already demonstrated the economics of near-zero marginal reproduction. Once software exists, producing another copy costs almost nothing.4647AI potentially extends this property from **software itself to some of the work performed through software**.4849That is a much larger transformation.5051## The Strange Economics of Digital Workers5253Imagine a capable AI agent that costs $1 per hour to operate.5455Now imagine it improves until it performs certain tasks comparable to a worker costing $50 per hour.5657Companies would have an enormous incentive to substitute the expensive input with the inexpensive one wherever possible.5859But something even more unusual happens.6061The AI worker can be copied.6263A company does not have to choose between one AI worker and another.6465It can run 10.6667Or 1,000.6869Or 100,000.7071The constraint eventually becomes compute rather than the availability of trained humans.7273This creates an economy where cognitive capacity could become elastic in a way human labor never was.7475If a problem requires ten times more analysis, you allocate ten times more compute.7677If a software project benefits from hundreds of parallel experiments, you launch hundreds of agents.7879If a scientific question requires reading one million documents, machines can divide the corpus among thousands of instances.8081This is not merely automation.8283It is the potential industrialization of cognition.8485## Cheap Intelligence Does Not Mean Cheap Everything8687There is an important mistake in assuming that cheaper intelligence makes everything abundant.8889It does not.9091Some resources remain fundamentally scarce.9293Land is scarce.9495Energy infrastructure is scarce.9697Advanced semiconductor manufacturing capacity is scarce.9899Certain minerals are scarce.100101Physical transportation has constraints.102103Housing in desirable locations is scarce.104105Human attention is scarce.106107Time itself remains scarce.108109As intelligence becomes cheaper, these complementary scarce resources may actually become **more valuable**.110111This pattern has occurred before.112113When one factor of production becomes dramatically more abundant, economic value often migrates toward the remaining bottlenecks.114115Cheap computation increased the importance of data.116117The internet made information abundant while increasing the value of attention.118119AI could make cognitive production abundant while increasing the value of energy, compute infrastructure, proprietary data, physical assets, distribution, trust, and human attention.120121The interesting question therefore isn't simply:122123> What will AI make cheaper?124125It is also:126127> What becomes more valuable because intelligence became cheaper?128129## Capital May Become More Important130131This leads to an uncomfortable economic possibility.132133AI is often discussed primarily as a labor technology.134135It may be more useful to think about it as a new form of capital.136137A machine, server, model, or agent can perform work repeatedly after an initial investment.138139If cognitive production becomes increasingly capital-intensive, ownership of productive assets becomes more important.140141Consider two individuals.142143One primarily earns income by selling cognitive labor.144145The other owns compute infrastructure, models, software, businesses, energy assets, or financial capital.146147If machines increasingly substitute for cognitive labor, their economic positions may diverge.148149The first individual competes with increasingly inexpensive machine intelligence.150151The second owns assets whose productivity is amplified by that intelligence.152153AI could therefore create extraordinary productivity growth while simultaneously increasing the importance of capital ownership.154155Those outcomes are not contradictory.156157## The One-Person Corporation158159There is another side to the same phenomenon.160161Historically, building a large company required coordinating many specialized people.162163Engineering.164165Accounting.166167Marketing.168169Customer support.170171Legal work.172173Research.174175Operations.176177Management.178179Each function created organizational complexity.180181AI agents could dramatically reduce the minimum human organization required to operate a sophisticated company.182183Imagine one entrepreneur coordinating dozens or hundreds of specialized agents.184185One agent maintains infrastructure.186187Another analyzes customer feedback.188189Another generates software tests.190191Another monitors financial metrics.192193Another conducts market research.194195Others handle documentation, localization, analytics, or internal operations.196197The human becomes less of a worker performing every task and more of an allocator of machine intelligence.198199This could produce something historically unusual:200201**extremely small organizations controlling extremely large amounts of productive capacity.**202203The famous one-person billion-dollar company may or may not appear soon.204205The more important observation is that the minimum number of humans required to operate a given amount of economic activity is likely to decline.206207That alone could substantially reshape firms.208209## Science Could Experience the Same Transformation210211The economics of research are particularly interesting.212213Scientific progress is partly constrained by the cost of experimentation.214215Researchers must search literature, clean datasets, write software, formulate models, conduct robustness tests, analyze results, and document experiments.216217Much of this work is cognitive and computational.218219Agents could make experiments dramatically cheaper.220221Instead of asking:222223> Which three specifications should we test?224225A researcher might eventually ask:226227> Which 30,000 specifications should the system explore, and how should we statistically evaluate the resulting evidence?228229The scarce resource moves upward.230231Execution becomes cheap.232233Choosing meaningful questions becomes valuable.234235This could produce a paradoxical effect: as machines become better at research tasks, human scientific judgment may become more important rather than less.236237Generating experiments is not the same thing as knowing which experiments matter.238239## Software Is the First Laboratory240241Software development provides perhaps the clearest preview.242243Programming is unusually compatible with AI because the environment provides rapid feedback.244245The model writes code.246247The computer executes it.248249The program fails.250251The model reads the error.252253It modifies the code.254255The program runs again.256257This feedback loop allows AI systems to move beyond generating text toward actually completing tasks.258259The same architecture can eventually extend elsewhere whenever environments provide machine-readable feedback.260261Finance.262263Engineering.264265Scientific computing.266267Logistics.268269Accounting.270271Design.272273Operations.274275The deeper transformation therefore comes not simply from better language models, but from connecting models to environments where they can **act, observe results, and iterate**.276277That is the transition from models to agents.278279## Productivity Could Become Difficult to Measure280281If this transition occurs quickly, traditional measures of economic activity may initially struggle to capture it.282283Suppose an individual uses local AI agents to produce software that would previously have required a ten-person company.284285The amount of economically useful output may increase enormously without a proportional increase in wages, employment, or organizational size.286287Similarly, open-source models running on privately owned hardware can generate valuable output without creating an API transaction every time intelligence is consumed.288289Some intelligence may become an internal intermediate good.290291A company could consume billions of machine-generated tokens internally to optimize operations while very little of that activity appears directly as final expenditure.292293Economic statistics were designed around economies where production largely involved observable transactions between humans and organizations.294295Machine-generated internal cognitive production could complicate that picture.296297## Intelligence Will Still Have a Price298299There is a final qualification.300301The cost of intelligence probably never reaches zero.302303As inexpensive models become capable enough for ordinary tasks, demand for more capable intelligence may expand.304305A cheap model might solve a problem in one second.306307A more sophisticated system might spend one thousand times more computation searching for a significantly better solution.308309This resembles computing itself.310311Computers became dramatically cheaper.312313Humanity did not respond by spending less on computation.314315We found vastly more things to compute.316317The same may happen with intelligence.318319As the price per unit falls, civilization may consume extraordinary quantities of it.320321Companies could run continuous simulations.322323Scientists could launch millions of experiments.324325Software agents could continuously optimize infrastructure.326327Individuals could maintain personalized models analyzing enormous amounts of information.328329The total amount spent on machine intelligence could therefore increase even while its unit cost collapses.330331## The Real Question332333The most interesting future is not necessarily one where artificial intelligence replaces humans.334335It is one where intelligence becomes an abundant industrial resource.336337That would force the economy to reorganize around whatever remains scarce.338339Compute.340341Energy.342343Physical resources.344345Capital.346347Trust.348349Attention.350351Ownership.352353And, perhaps most importantly, good questions.354355For centuries, civilization operated under a fundamental constraint: meaningful cognitive work required human cognitive time.356357We built companies, universities, governments, and markets around that constraint.358359If machine intelligence significantly weakens it, we should expect more than productivity improvements.360361We should expect new organizational structures, new distributions of economic power, new business models, and possibly entirely new ways of conducting science.362363The important question may therefore not be whether artificial intelligence becomes smarter than humans.364365It may be much simpler:366367**What does an economy look like when intelligence is no longer scarce?**368369