翻訳待ち：The AI Apocalypse Is Here

AI サービスが一時的に利用できないため、復旧後に翻訳を補完します。

Americans do not care for artificial intelligence. Recent polling shows that their attitudes mostly range from ambivalence to horror. Some observers have suggested that opposition to AI is due to poor PR from industry leaders. If tech titans hadn’t been so convinced that boasting of their apocalyptic powers was the best way to raise capital, the argument goes, then the American public would be on board. Chinese public opinion, for instance, is considerably more favorable. But these discrepancies can be read another way. The taste for liberty is not independent of culture and history. China has not known freedom for some time, but citizens of the United States have a long experience of it. It is a testament to Americans’ retention of the spirit of liberty that they mistrust AI so much. The various gripes against AI are manifestations of a largely inarticulate intuition that its widespread adoption cannot be reconciled with a free society. Americans are groping toward the realization that, even if AI does not end life on earth, it will destroy our way of life. In order to make the case against AI, one need only quote the industry’s leaders. They have foretold the creation of a “permanent underclass” once AI agents (imminently, they think) surpass us in all cognitive capacities, and have long warned that misaligned AI could even lead to the extinction of humanity. This is a far cry from anything America has seen before from its captains of industry. Andrew Carnegie never promised to bring on the apocalypse. While industry leaders have softened their rhetoric of late, now disavowing the goal of complete human economic obsolescence, the major AI corporations remain officially committed to the development of artificial general intelligence, defined as an AI system that “outperforms humans at most economically valuable work.” Already, AI is outpacing what humans can do in many research areas, and AI-generated text and music are overwhelming the digital landscape. It is impossible to know what will actually come to pass. But given the extraordinary pace at which the technology is progressing, it would be foolish to underrate AI’s disruptive capacity. The present moment is awash in schemes for regulating AI, some in radical ways. Plans for partial nationalization of the AI companies, endorsed by figures from right-populist guru Steve Bannon to leftist hero Sen. Bernie Sanders (I.-Vt.); Sen. Marsha Blackburn’s (R.-Tenn.) sketch of regulations that will protect “children, creators, conservatives, and communities”; Rep. Alexandria Ocasio-Cortez’s (D.-NY) proposed moratorium on data center construction; the Trump administration’s recent interventions to halt the distribution of new models deemed to jeopardize national security; the escalating calls for an international treaty to ban superintelligence—these are all salutary signs of an awakening to the threats that AI poses. And yet even these critics do not seem fully to reckon with the fact that AI portends not merely adverse economic or security impacts, but an irrevocable change in human life. Even now, AI is having deleterious effects that legislation to redistribute income or transfer control of the technology will do little to address. In particular, the most vocal skeptics have not fully reckoned with how anthropomorphic AI—models that mimic human personality traits, emotional states, and patterns of thought, as currently commercially available generative AI models do—is currently in the process of unsettling human psychology and the social fabric. “There are also good reasons to oppose generative AI as we already know it.” In other words, the present AI resistance speaks almost entirely of preventing risks still to come. But there are also good reasons to oppose generative AI as we already know it. It is necessary to oppose not just what AI may become and will likely do to the country, but what it is now and is already doing to our society. Given what we have already seen of AI, we can be confident that its widespread adoption will create a less free society, a less virtuous people, and a degraded culture. A movement recognizing these facts would not call merely for caution and regulation, but for limiting the diffusion throughout civil society of anthropomorphic AI that competes with human authors and artists in creating new text, audio, and images, and for ending the pursuit of superintelligence. So far, opponents of AI have overwhelmingly focused on predicting its economic effects. But AI is more than just another industrial revolution. It is—already—a moral and cosmological cataclysm. For generative AI has done nothing less than change the place of human beings in the cosmos. More than the printing press or the internet, perhaps more than anything since the rise of agriculture, AI alters the relationship of human beings to their world by undermining our status as the distinctive bearers of language. While Bernie Sanders and Ron DeSantis have both pointed to the loss of control of our destiny as one of AI’s great risks, no leader has yet reckoned with how astonishing it is that the only form of general intelligence in our world is no longer human intelligence, nor contended with the truth that widespread interaction with machines endowed with conversational language abilities and superhuman cognitive power is the greatest social-psychological experiment the species has ever run. AI skeptics by and large have not focused on the loss of human distinctiveness that large language models have brought about, nor have they gamed out all the ways in which, even if (an unlikely proposition) “safety” could be guaranteed, moral and cultural harms will ensue from its rise. AI should worry those of all ideological stripes, but it is notable how muted conservatives have been on the subject. For not merely the prospect of superintelligence, but also the present widespread deployment of anthropomorphic AI, threaten the things that conservatives profess to hold dearest: capitalism, the work ethic, personal responsibility, individual freedom, and common-sense moral standards. Although Americans have not looked kindly on the rise of AI, no organized movement has yet developed against it. One reason for this, perhaps, is the cult of the new, which is the closest thing to an American civic religion that still exists. This penchant for exaltation of novelties is, curiously, reinforced by an equally powerful tendency to deny that there is anything new under the sun. Libertarian-leaning thinkers in particular defend their embrace of anthropomorphic AI by deriding critics as Luddites and chiding them for giving off the “same energy” as people who shut down nuclear power plants and would have resisted the steam engine had they been alive for the industrial revolution. But there are, in fact, new things under the sun, and generative AI is one of them. Critics of AI are not arguing against the latest auto transmission or better solar panels. There is a difference in kind between a technology that makes material processes more efficient in some discrete domain of activity (as previous innovations did), and one expected to surpass all humans in creative and cognitive capacity. The line of defense that likens AI to prior industrial advances and insists it is a “normal technology” requires adopting a patronizing attitude toward the industry’s own leaders, who do not believe that they are merely bringing about a new industrial revolution, but who instead think they are developing a technology that will cause an even more far-reaching set of societal changes and indeed even alter (or displace) human nature itself. The implicit suggestion is that the reason we should take a hands-off approach to AI is that the sector’s own leading figures are so deluded as to misunderstand what they are up to. Claims based on analogies to the history of economic innovation function as a genre of moral blackmail, an insinuation that having objections to AI is equivalent to wanting to give up on modern medicine or mass production. Fortunately, most Americans are able to make common-sense distinctions. That we managed to adapt in the past (often with much greater hardship than we now recognize) to railroads and factories and telephones gives us no reason to be sanguine about the effects of this unique technology on our society. Instead of deferring to novelty, it is our duty as citizens to evaluate new technologies with as much skepticism as we do new policy proposals. Perhaps the most astonishing technical feat of modern bioscience is the development of the capacity to clone humans, but regulations and social conventions severely limit the practice. We could increase GDP by rescinding environmental standards of a century ago or by relegalizing child labor, but we rightly do neither. There are more goods in life than increased productive capacity, and there’s no more reason to imagine that every technological development or economic initiative will benefit us than to think that every new political or moral idea is correct. There is no way around judging the value of each innovation for its own sake. That is what democratic citizenship demands of us. Unfortunately, AI-critical argumentation has been flawed. The same progressives who have led the charge against AI have recently embraced degrowth ideas, and have even held up the very idea of economic progress as a pernicious myth. Claiming that AI will consume too many resources or cause unacceptable environmental harm reflects this tendency—and is empirically overblown. Likewise, progressives have expressed their concerns about AI in an anti-capitalist idiom. All of this makes it easy for AI apologists to portray any critique of AI as issuing from a general hostility to business success and an insatiable appetite for regulation. To most Americans, this mindset is perverse and alienating. What is needed is an anti-AI movement that accords with broadly held American values. “Sanders and others are right to frame AI as endangering democracy.” It would be wrong, though, to be too hard on progressives. They are coming to the fore, and they do have a few resonant themes to play upon. For years now, progressivism has conflated ordinary policy disagreements with threats to democracy. Here, however, their diagnosis is apt. Sanders and others are right to frame AI as endangering democracy. Democracy simply cannot survive if the economic contributions of most people are rendered irrelevant, an eventuality which AI leaders foresee and have often seemed to revel in. In a world of robot soldiers of the sort some AI champions envision, in which the willingness of the average citizen to fight and die for his country is no longer necessary to the well-being of the state and national elites, there will be much less reason for governments to accord citizens rights. Along similar lines, artificial general intelligence creates a technological version of the “resource curse”—the theory that abundant natural resources can impede countries from democratizing and modernizing because they provide a source of revenue independent of the need to tax productive, laboring citizens. If citizens no longer have economic or military value, they will become subjects. If AI swallows the rest of the economy, then either the AI companies will directly rule us, supplanting the traditional state (a prospect about which some of their promoters fantasize), or the state will be able to extract more revenue from the companies than from taxing citizens. In either case, normal people will have less of a claim on the concern of those who govern them. Moreover, while Tech Right leaders such as Elon Musk used to complain about censorship and cancel culture, it’s clear already that AI is a far more powerful technology for the suppression of unwanted expression than social media algorithms. Its ability to identify and track sp [truncated for AI cost control]