The Right Needs AI Realism
Neither luddism nor accelerationism is a serious policy.
The right is finally having its long-overdue AI fight.
Conservatives have treated AI as either a productivity miracle or a distant science-fiction threat. But the coalition lines are not clean. Parts of Silicon Valley now live comfortably inside the right, and parts of the right have adopted the language of founders, defense tech, crypto, and acceleration. The boosters tell us to build faster, regulate less, and trust the engineers. The skeptics warn of job loss, censorship, surveillance, and machines that slowly displace human judgment. Both camps see something real. Neither offers enough.
The recent conservative backlash against the administration’s AI policy shows that the debate has entered a new phase. More than 60 Trump allies urged him to require testing and approval of powerful AI models before release. That put one vocal faction of the MAGA base at odds with the White House’s more permissive approach. The administration had been preparing an AI order that would have created a voluntary framework for labs to give the government access to covered frontier models before public release. Then Trump called it off on May 21, saying he did not want to do anything that could dull America’s AI edge. That reversal only sharpens the question: Who governs the machines?
That question is the beginning of AI realism.
A conservative politics of AI cannot be written solely by people whose first loyalty is valuation, scale, data capture, or market dominance, even when those people now sit inside the conservative coalition. But neither can it be written by people who imagine the machine age can be wished away. The right should reject both uncritical acceleration and reactionary panic. Artificial intelligence is not merely another innovation story, labor story, censorship story, or China competition story. It is becoming a layer between citizens and reality itself.
AI systems increasingly shape what people search, what they read, what they believe is credible; they determine what work remains human, how students learn, how bureaucracies make decisions, how campaigns test persuasion, and how institutions understand the public. The old public square was noisy, biased, and often foolish, but at least citizens could see much of the machinery. The new public square is filtered, ranked, summarized, personalized, and increasingly generated. The danger is not that every answer will be false. The danger is that reality becomes too mediated for ordinary citizens to know where persuasion ends and judgment begins.
That makes AI a problem of self-government. Conservatives should be especially alert to this. A republic depends on citizens capable of judgment. It depends on institutions that remain accountable to the people they serve. It depends on some shared capacity to distinguish fact from fiction, speech from manipulation, authority from expertise, and public power from private preference. AI touches all of these. If search engines, chatbots, recommendation systems, automated agents, and synthetic media become the primary interface through which citizens encounter the world, then AI policy becomes civic policy.
The fight is no longer theoretical. The U.S. government’s dispute with Anthropic shows what happens when private AI vendors, national-security agencies, and procurement power collide. The administration has defended its designation of Anthropic as a supply-chain risk even while exploring ways to use the company’s powerful Mythos model against cyber threats. The Pentagon argues it cannot rely on a vendor that might pull the plug over its AI safety views. Anthropic says it cannot control models once deployed in classified settings and has drawn red lines around mass domestic surveillance and autonomous weapons.
That is not a narrow contracting dispute. It is a preview of AI governance. Who decides the limits once these tools become part of public power: elected officials, private model providers, courts, procurement officers, or corporate safety teams?
The answer cannot be that a handful of private AI firms quietly decide the boundaries of public authority. But the answer also cannot be that the state absorbs every frontier model into its own machinery without meaningful limits. In a healthy republic, national-security tools should answer to constitutional government. So should the agencies that deploy them. Private vendors should not possess veto power over the state, but the state should not use procurement as a way to erase ethical limits, contractual boundaries, or public accountability.
AI also has a physical component. It needs land, water, electricity, transmission lines, tax deals, and local political permission. The proposed Stratos project in Box Elder County, Utah, would bring large-scale data centers and energy development to about 40,000 acres, while critics have questioned state-level review, water oversight, and who has the final say. In Cape Town, Equinix faces opposition from groups demanding fuller disclosure of water, power, and environmental impacts before local planners approve two major data centers. The machine age is not floating in the cloud. It is landing in counties, towns, grids, aquifers, and neighborhoods.
A politics that speaks endlessly of innovation while ignoring ratepayers, residents, towns, counties, and local governments is not conservative. It is abstraction with venture funding.
This is why the technological class cannot become the right’s governing class by default. The men who build the machines should be heard. Some of them are genuine allies. Many are patriotic, brilliant, and necessary. But they should not be enthroned.
Technology companies are good at building tools. So are the founders, investors, engineers, and defense-tech entrepreneurs who increasingly shape the right’s imagination of the future. They are not wrong to want American strength. They are often right about stagnation, bureaucracy, and the need to build. But the habits that produce software and scale companies are not enough to answer political questions about family, place, nation, duty, citizenship, moral formation, public trust, or constitutional order.
Their incentives are speed, scale, user growth, investor confidence, and market capture. Those incentives can produce astonishing things. They can also produce systems that reshape civic life long before citizens understand what has happened.
The answer is not for the right to turn against its builders. Conservatives should remember that fear of technological displacement is not new. The machine has frightened workers, writers, politicians, and philosophers since the first industrial revolution. The original Luddites were not stupid men smashing progress because they hated tools. They were skilled workers responding to economic disruption, social displacement, and loss of control. Their anxiety was human. Their strategy was not enough.
The same is true now. The answer is not to rage against the machine. The answer is to ask who owns it, who governs it, who benefits from it, and whether citizens remain capable of saying no.
A conservative AI realism should begin with innovation, but not end there. America should build. Retreating from AI would hand the future to rivals who do not share our constitutional traditions or concern for human liberty. National strength requires technological strength. But strength without judgment becomes appetite. The point is not merely to win the AI race. The point is to remain the kind of civilization worth winning it for.
That means preserving human judgment in high-stakes decisions. It means refusing to let automated systems replace accountable officials in contexts where law, liberty, life, livelihood, or civic standing are at issue. It means protecting speech while recognizing that automated persuasion, bot networks, deepfakes, and synthetic media can distort the conditions under which free speech is understood. It means demanding auditability where AI systems exercise public power. It means preserving model replaceability and public control when private tools enter government workflows. It means building energy and data infrastructure with respect for towns, residents, ratepayers, counties, and local consent rather than treating them as obstacles to scale.
It also means taking literacy seriously. AI literacy is civic literacy now. Citizens do not all need to become machine-learning engineers. But they do need to know when a system is helping them think and when it is laundering someone else’s assumptions. They need to understand that chatbots have defaults, models have training histories, search has rankings, feeds have incentives, and every interface teaches habits of mind.
This is not a call for a new priesthood of regulators. It is a call for political maturity. The right should resist the temptation to solve the AI question with a centralized speech regime, safety bureaucracy, or industrial policy written by the loudest incumbent firms. But it should also resist the fantasy that markets alone can answer civilizational questions. Markets are good at discovering price. They are not sufficient for preserving a republic.
The best future for AI is not one in which machines replace human judgment, but one in which they strengthen the conditions under which human judgment can flourish. Used well, AI could help citizens understand complex policy, help small businesses compete, help doctors and teachers extend their judgment, and help local governments become more competent. It could help America remain technologically sovereign in a dangerous world.
Used badly, AI becomes a machinery of dependency. Citizens lose the ability to tell what is real. Workers are managed by invisible systems. Schools outsource judgment. Agencies hide behind automation. Private vendors quietly mediate public life.
AI realism is therefore not pessimism. It is technology ordered toward human ends: capable citizens, stronger communities, better work, truthful information, accountable institutions, national resilience, and public power that remains answerable to the people.
The right should not ask whether AI can be stopped. It should ask what kind of future AI is being used to build. The machine age will not wait for conservatives to settle their theory of technology. But a republic that still believes in human dignity, ordered liberty, and self-government has something better to offer than panic or surrender. It can insist that intelligence, artificial or otherwise, be placed in the service of human flourishing.
The post The Right Needs AI Realism appeared first on The American Conservative.
