AIがもたらす大量破壊型資本主義

Extinction-level capitalism a citizen’s thoughtson AI riskAI is inher­ently polit­ical tech­nology. If AI works as intended, it will grad­u­ally corrode our liberal democ­racy, risking an irre­versible shift into another polit­ical and economic config­u­ra­tion. Among AI risks, this one deserves more consid­er­a­tion, because it requires no addi­tional condi­tions like malign actors or AI malfunc­tion. AI only needs to amplify existing trends, espe­cially around concen­tra­tion of capital. This damage will occur even assuming that in the near term, AI will broadly improve mate­rial well-being.About MBI’m a self-employed author, designer, programmer, and lawyer. In 2022, I learned that my own works were in the training datasets of gener­a­tive-AI compa­nies. In response, I invented the first set of lawsuits chal­lenging the legality of these prac­tices. I’m currently co-counsel for plain­tiffs in a number of AI cases. Though I discuss certain legal issues below, I am not your lawyer, and nothing here is held out as legal advice. These are my personal views as a citizen and economic actor; I speak only for myself. This piece is typeset in Equity, Advo­cate, and Trip­li­cate, fonts I designed. They can be licensed for your own polemics and pamphlets.Emergent effectsTwo billion years ago, the rock layers comprising what is now called the Colorado Plateau began to form: first igneous and meta­mor­phic rocks, followed by many layers of sedi­men­tary rocks. About fifty million years ago, through tectonic action, this plateau gained thou­sands of feet of eleva­tion. About five million years ago, a river began to flow. The river carried silt and debris, scraping out the begin­nings of a canyon. The river deep­ened the canyon, exposing its walls to weather and erosional forces that widened the canyon further. Today the waterway is the Colorado River. The geolog­ical forma­tion is the Grand Canyon.The forma­tion of the Grand Canyon required zero human agency. Zero tech­nology. Zero coor­di­na­tion among the river, the land, and gravity. In that sense the Grand Canyon is an emer­gent effect: a complex, unfore­see­able output arising from simpler inputs.But we would never wonder whether the river is sentient. Or whether the river cares about the dirt that it carries out of the canyon. The water is just doing what water does: flowing down­hill. The dirt just happens to be in the way.Inherently political technologyLangdon Winner is a polit­ical theo­rist. Winner wrote the excel­lent and influ­en­tial essay “Do Arti­facts Have Poli­tics?” (1980). Winner sought to debunk the tradi­tional framing that “tech­nolo­gies are … neutral tools that can be used well or poorly, for good, evil, or some­thing in between.” Instead, Winner proposes two ways that a tech­nology can affect its polit­ical envi­ron­ment:The tech­nology is designed to have certain polit­ical effects. For example, the Great Fire­wall of China, a bundle of tech­no­log­ical measures that limit Chinese citi­zens’ access to foreign infor­ma­tion sources. Antipo­dally, the Tor Project intends to maxi­mize user anonymity and thwart govern­ment intru­sion.The tech­nology is inher­ently polit­ical. This is Winner’s key analytic fulcrum. Winner describes two versions of inher­ently polit­ical tech­nology. The first is where the tech­nology “actu­ally requires … a partic­ular set of social condi­tions as [its] oper­ating envi­ron­ment.” For instance, nuclear weapons: the only respon­sible way to possess such dangerous tech­nology is to place it within “a central­ized, rigidly hier­ar­chical chain of command … the [atom] bomb must be author­i­tarian; there is no other way.” The second version is where the tech­nology is “strongly compat­ible” with a certain polit­ical arrange­ment (even if not strictly required) and thus tends to bring that arrange­ment to fruition.As an example, Winner considers the mechan­ical tomato harvester. Devel­oped at UC Davis in the 1950s, the machine was tremen­dously produc­tive. But it was also expen­sive. Only well-capi­tal­ized tomato growers could afford it. The rest couldn’t compete. According to Winner, the number of Cali­fornia tomato growers dropped from ~4000 in the early 1960s to ~600 in 1973, costing ~32,000 jobs and the compounding nega­tive effects on those commu­ni­ties. Winner summa­rizes:What we see here … is an ongoing social process in which scien­tific knowl­edge, tech­no­log­ical inven­tion, and corpo­rate profit rein­force each other in deeply entrenched patterns that bear the unmis­tak­able stamp of polit­ical and economic power … oppo­nents of inno­va­tions like the tomato harvester are made to seem “antitech­nology” or “antiprogress”. For the harvester is not merely the symbol of a social order that rewards some while punishing others; it is in a true sense an embod­i­ment of that order.Not merely the symbol—the embod­i­ment. A facially neutral tech­no­log­ical inven­tion—here, a tomato harvester—can induce polit­ical effects. Those effects don’t arise from flaws in the tech­nology. To the contrary—they arise from its effi­cacy.How are the polit­ical effects deter­mined? Winner iden­ti­fies two key early deci­sions. The first is the binary ques­tion of whether to pursue the tech­nology at all. The second are choices about “the design or arrange­ment” of the tech­nology. Winner cautions: “[t]o see the matter solely in terms of cost-cutting, effi­ciency, or the modern­iza­tion of equip­ment is to miss a deci­sive element”. That is, the polit­ical effects can possibly be coun­tered, but first they must be acknowl­edged.Of course, the best oppor­tu­nity to choose wisely is before the tech­nology is widely intro­duced, as capital and social invest­ment tends to entrench it:Because choices tend to become strongly fixed in mate­rial equip­ment, economic invest­ment, and social habit, the orig­inal flex­i­bility vanishes for all prac­tical purposes once the initial commit­ments are made. In that sense tech­no­log­ical inno­va­tions are similar to legisla­tive acts or polit­ical found­ings that estab­lish a frame­work for public order that will endure over many gener­a­tions. … The issues that divide or unite people in society are settled not only in the insti­tu­tions and prac­tices of poli­tics proper, but also, and less obvi­ously, in tangible arrange­ments of steel and concrete, wires and tran­sis­tors, nuts and bolts.Tech­no­log­ical choices bear directly on the “public order” at large. When we don’t take those choices seri­ously—or we’re persuaded to ignore them by those insisting that tech­nology is just a neutral tool—we risk polit­ical conse­quences.Winner warns of compla­cency. Once the tech­nology arrives and becomes entrenched, the conver­sa­tion gets reframed as one of tech­no­log­ical inevitabilism vs. anachro­nism, and dissent is discour­aged: “the kinds of reasoning that justify the adap­ta­tion of social life to tech­nical require­ments pop up as spon­ta­neously as flowers in the spring … After a certain point, those who cannot accept the hard require­ments and imper­a­tives will be dismissed as dreamers and fools.”Liberal democracyThe balance of power of democ­racy is premised on the average person having leverage through creating economic value. If that’s not present, I think things become kind of scary.—a certain AI CEOLiberal democ­racy is the polit­ical scien­tist’s term for the type of govern­ment preva­lent among capi­talist economies since the Amer­ican and French Revo­lu­tions. The intel­lec­tual foun­da­tion of liberal democ­racy arose during the Enlight­en­ment, espe­cially through the work of John Locke. Liberal democ­racy empha­sizes limited govern­ment, indi­vidual rights, and sepa­ra­tion of powers—in short, majority rule with excep­tions and guardrails. (The term liberal democ­racy doesn’t connote liberals or Democ­rats in the specific US polit­ical sense. But polit­ical parties of differing ideolo­gies are a tradi­tional feature of liberal democ­ra­cies.) Today, most liberal democ­ra­cies are in Europe, the Amer­icas, and the Pacific Rim.Liberal democ­racy is not a fixed set of immutable char­ac­ter­is­tics, but a bundle of graded values. All liberal democ­ra­cies empha­size certain ones over others. In the aggre­gate, some of these nations evolve toward stronger liber­alism; others evolve away. These degraded cases have some­times been called illib­eral democ­racy: the observ­able formal­i­ties of liberal democ­racy may still be observed—e.g. multi­party elec­tions, sepa­ra­tion of powers—but the lived reality is single-party rule and declining indi­vidual rights.That’s not to say that liberal democ­racy produces excel­lent outcomes for all citi­zens, all the time. It doesn’t. At any moment, certain citi­zens are dissat­is­fied—say, because they belong to a group whose rights are inad­e­quately protected or econom­i­cally margin­al­ized. Liberal democ­racy offers a process, not a result: grass­roots demo­c­ratic partic­i­pa­tion can coalesce into policy change. But within the arena of competing polit­ical inter­ests, winners and losers neces­sarily follow. Navi­gating these differ­ences within a stable, accom­moda­tive frame­work is prefer­able to a rigid one that buckles under these stresses—say, through polit­ical revo­lu­tion, which tends to be messy and unpre­dictable.Capi­talism has tradi­tion­ally been consid­ered a neces­sary but not suffi­cient condi­tion for liberal democ­racy. Why? A regu­lated market economy encour­ages citizen partic­i­pa­tion through prop­erty owner­ship and trans­ac­tion. A prin­cipal func­tion of govern­ment is to define the economic condi­tions of the state. Economic partic­i­pa­tion mutu­ally rein­forces demo­c­ratic partic­i­pa­tion. Prop­erty owners will vote for those who protect their inter­ests. The rise of indus­trial capi­talism in the 19th century, and the wealth-redis­tri­b­u­tion mech­a­nisms that followed in the 20th, led to economic empow­er­ment for more citi­zens, and ulti­mately broader polit­ical empow­er­ment. The converse also holds: economies premised on state or oligarchic control of some narrow class of assets haven’t tended to evolve toward liberal democ­racy.In prac­tice, certain people in a capi­talist liberal democ­racy tend to get increas­ingly rich. Absent coun­ter­mea­sures, the wealthy gain control of the polit­ical appa­ratus, thwarting liberal-demo­c­ratic norms. This tension between capital and poli­tics is a long-consid­ered topic. A key early work was, of course, Karl Marx’s Capital (about which more later). In the current era, Mancur Olson’s book The Rise and Decline of Nations set out how small groups with a shared interest (which could include capital concen­tra­tion) can effec­tively under­mine stable soci­eties. More recently, econ­o­mists Robert Reich (“How Capi­talism is Killing Democ­racy”), James Galbraith (The Predator State), and Yanis Varo­ufakis (Tech­nofeu­dalism: What Killed Capi­talism) are among those who have studied the esca­lating polit­ical conse­quences of rising wealth inequality. The synthesis might be: as more wealth becomes concen­trated in the hands of fewer citi­zens, liberal democ­racy weakens, because whichever citi­zens are losing economic rele­vance will also lose polit­ical rele­vance. A nation sending many of its citi­zens toward economic irrel­e­vance risks becoming polit­i­cally illib­eral.The Skynet fallacyAI discourse often invokes sci-fi narra­tives. I’ve called this the Skynet fallacy, after the Termi­nator antag­o­nist, the most cited. But any sci-fi will do. For instance, one AI CEO warned of AI “going Termi­nator”; Stephen Hawking and other scien­tists warned of AI “devel­oping weapons we cannot even under­stand”; a second AI CEO said we “don’t need much imag­i­na­tion [about AI risk] because we grew up with that in the media”; a third AI CEO invoked the sci-fi movies Contact and 2001: A Space Odyssey in a piece about AI risk; a US congressman summa­rized AI risk as “evil robots rising up to take over the world”; a well-known jour­nalist advo­cated for more Termi­nator analo­gies; a promi­nent AI-risk pundit said he’s “annoyed” with Termi­nator analo­gies yet has suggested that AI will erad­i­cate humanity using hordes of toxic nanobots.Artis­ti­cally, sci-fi movies exter­nalize the awe and unease of tech­no­log­ical confronta­tion. It’s easy to see why these metaphors have become part of AI-risk discourse. And yet. As AI puts down roots in our economy, the sci-fi framing hides more than it reveals. Sci-fi plots are opti­mized for cine­matic impact. So as a metaphor for AI risk, they can lead to faulty intu­itions. Among real­istic AI risks, we can expect that most will be boring, slow, and depend on minimal extra tech­nology. Whether AI will cause literal human extinc­tion is esoteric—a light­ning strike. But AI could easily induce future economic and polit­ical condi­tions that most Amer­i­cans today would consider intol­er­able—a cancer that extin­guishes a certain way of life. Nobody’s going to make a movie about boring AI risks. But they comprise the majority of worri­some AI outcomes.In 2003, philoso­pher Nick Bostrom proposed his now well-known parable of the paper­clip maxi­mizer. Bostrom imag­ines an advanced AI that is asked to make paper­clips. Taking its mission seri­ously, the AI erad­i­cates humanity as it consumes all Earth resources to make more paper­clips. Bostrom was illus­trating the AI control problem: ensuring an AI acts consis­tently with human prior­i­ties is diffi­cult, even for osten­sibly simple goals. Because when we say “AI, make paper­clips”, the implied coda is “… without killing everyone.” But an AI can’t know this ex nihilo. Bostrom softens the sci-fi flavor by choosing paper­clips and not, say, laser-wielding robots. Bostrom’s choice of an economic mech­a­nism of resource conver­sion is apt. Even a mundane objec­tive can produce outsize risk. We could further observe that on the paper­clip-maxi­mizing path, human life would become dystopic long before literal extinc­tion. As resources are depleted, humans would become tenants in a neofeu­dalist paper­clip empire. (Paper­clip Crisis: The Saga Begins—opening soon.)Computer scien­tist Stuart Russell also explored the control problem in his book Human Compat­ible. Russell calls one variant the gorilla problem: that “ances­tors of the modern gorilla created … the genetic lineage leading to modern humans. How do the gorillas feel about this? … the consensus opinion would be very nega­tive indeed.” Despite shared lineage, gorillas and humans have incom­men­su­rable values. Nothing humans can do—short of disap­pearing—would restore gorillas to their golden age. Russell’s framing is a believ­able analogy for the future rela­tion­ship of humans and AI. Sure, gorillas didn’t delib­er­ately invent humans. But I take Russell to mean that the emer­gent char­ac­ter­is­tics of these rela­tion­ships are more conse­quen­tial than the intended ones. (In that sense, Russell echoes Langdon Winner.) The fact that we’re inventing AI doesn’t mean we will predict or control its gravest effects. Any more than gorillas could predict or control human domi­nance of their ecosystem. The gorillas did their thing. We did ours.AI will do its thing too. It will take time to figure out what that is, exactly.How the West was wonI recently read Cadillac Desert by Marc Reisner, about the devel­op­ment of water resources in the western US between 1910 and 1980, espe­cially the dam-building campaign of the federal Bureau of Recla­ma­tion. Reisner’s book weaves several story­lines:Engi­neering and envi­ron­mental impacts. Dams concen­trate water. But they cause envi­ron­mental conse­quences else­where. Further­more, Recla­ma­tion’s 20th-century projects were often based on opti­mistic projec­tions of mete­o­ro­log­ical water supply. Today, long-term drought condi­tions chal­lenge those projec­tions. No amount of money or hydro­logic engi­neering can change that.National US poli­tics. In the early 20th century, newer Western states sought polit­ical clout in the federal govern­ment, which had been domi­nated by Eastern states. Polit­ical clout followed economic growth, and to achieve growth, the Western states depended on one crit­ical but scarce input: water. In the West, the federal Bureau of Recla­ma­tion was tasked with increasing water supplies. A polit­ical symbiosis emerged: congressmen from Western states voted to fund Recla­ma­tion water projects; in turn, Recla­ma­tion looked out for their constituents. Over decades, Reisner depicts this rela­tion­ship as metas­ta­sizing from prac­ti­cality to corrup­tion, in the sense of Recla­ma­tion becoming beholden to a narrow polit­ical lane. In that sense, Recla­ma­tion’s dams arguably qualify as inher­ently polit­ical tech­nology.State economies. Recla­ma­tion’s highly subsi­dized water boot­strapped Western economies, espe­cially agri­cul­ture. For a while, it worked as promised: Farmers got irri­ga­tion. Cities got water. Western states grew and pros­pered. But the projects worked so well that Western states wanted more. These states never weaned them­selves from subsi­dized federal water, setting them on a path toward perma­nent depen­dence.The parallel between water and AI is inexact. Water is a biolog­ical neces­sity; AI is not. Recla­ma­tion’s projects worked (up to a point); AI may or may not. This is part of why AI propo­nents have sought to raise the stakes. So far, AI has been grue­somely expen­sive and deliv­ered middling results. Never­the­less it’s routinely depicted as a geopo­lit­ical fulcrum, a proxy for contin­uing US excep­tion­alism. If Amer­i­cans don’t adopt AI whole­heart­edly, we will be losers. Do you want to be a loser?Labor replacementQ: What is Big AI primarily selling? A: Labor replace­ment, with mass unem­ploy­ment as a likely conse­quence. Some disagree or call it doomerish. Why? It’s exactly what AI grandees have been telling us. A certain AI CEO wrote that AI “will be hugely desta­bi­lizing for hundreds of millions” and that AI tools “are funda­men­tally labor replacing”. A certain AI company released a research paper about “the labor market impact poten­tial of large language models”. That AI CEO said “jobs are defi­nitely going to go away, full stop”. A second AI CEO said that in the near future, “20% of people don’t have jobs.” A third AI CEO predicted that farther out, “prob­ably none of us will have a job.” An AI-adja­cent CEO said that AI “will destroy human­i­ties jobs”. The ball is not hidden.Capital markets are already pricing in these expec­ta­tions. Regard­less of whether Big AI even­tu­ally delivers mass labor replace­ment, today these compa­nies seek to concen­trate capital as if they will. According to the Wash­ington Post, AI capital expen­di­ture in 2026 is esti­mated to be $700 billion, a “spending spree [that] has few prece­dents”. Based on a recent survey of US workers, the Global Part­ner­ship on Arti­fi­cial Intel­li­gence said that “the policy window to shape how AI trans­forms work is prob­ably closing faster than most govern­ments realize.”Extra­or­di­nary invest­ment demands extra­or­di­nary returns. In early 2026, after a certain billion­aire tech CEO laid off 40% of his employees and attrib­uted the deci­sion to a new “core thesis” of AI, the company’s stock rose nearly 25%. He won’t be the last. Whether these layoffs are based on actual AI bene­fits or merely “antic­i­pa­tory” is neither here nor there. Employers have strong incen­tives to reduce head­count and increase AI spending before competi­tors do. We will increas­ingly see both kinds of layoffs. Soft­ware program­mers are one set of conse­quen­tial, highly paid writers likely to be replaced with AI. Else­where I’ve predicted that legal prac­tice will also be seri­ously impaired. Why? Like program­mers, lawyers write about conse­quen­tial things and charge a lot. (We can expect th

※ 著作権に配慮し、引用は冒頭3段落までです。続きは元記事をご覧ください。