Earlier this year, in a conversation with the Wall Street Journal‘s editor-in-chief Emma Tucker at Davos, Dario Amodei said this:
We’re going to find that ideology will not survive the nature of this technology. It won’t survive reality.
He was talking about labor policy under advanced AI. What he was predicting is that the disruption will be so severe that support for wage insurance, retraining, and long-term income support will “become bipartisan and universal because everyone will recognize the necessity of it. Just mark my words.”
In an essay published a few months earlier, he made the geopolitical version of the same argument, writing that “just as feudalism became unworkable with the industrial revolution, the AI age could lead inevitably and logically to the conclusion that democracy... is the only viable form of government.” The claim in both cases is structurally the same. Politics is going to resolve itself into consensus because the material situation will force the answer.
The shape has a name in philosophy of science. It has been called scientism — the move of taking the authority that properly belongs to empirical inquiry into how the world works and applying it to questions about how human beings should live together, which are not questions empirical inquiry can settle. Physics can tell you what a object falling will do. What it will not do, is tell you whether wage insurance is the right response to displacement, or whether authoritarian consolidation is a historical inevitability, or whose interests a political settlement should serve. Those are value questions in the disguise of technical ones. So when someone such as Amodei says ideology will not survive reality, he is performing the scientism move at its cleanest: he is claiming that a political question — what should we do about mass technological unemployment, and on whose behalf — has a right answer that reality itself will impose, the way reality imposes the fall of a stone. It does not. No answer of that kind is waiting to be discovered. What exists is the settlement a particular constellation of powerful actors is preparing to impose, and the alternative settlements that could be fought for instead.
When people say they are post-ideological, what it is that they usually mean is they do not hold any of the packaged commitments that map onto the current tribal groupings. They will be very clear with you they are not Republican or a Democrat, that they aren’t a socialist or a libertarian, and they aren’t a progressive or a conservative. They have, you see, looked at the parties and found none of them acceptable, and they have concluded that the correct thing to do is exit the ideological game entirely and simply reason from first principles toward the best policy.
Fine. That’s a defensible pose to strike about the parties. It is not a defensible pose to strike about ideology. Because ideology is not the packaged set of party commitments. Ideology is the framework that decides what counts as a problem, what counts as a solution, whose interests are visible in the analysis, and what the word necessity is doing when you use it. You do not get to opt out of that. You can pretend you have opted out, but the pretense is itself an ideological move.
Ideology is not a bad word. Ideology is what happens when human beings live together. I am, for instance, extremely ideological about the proposition that murder should be prohibited and punished. I hold this ideologically. I did not arrive at it through some neutral procedure of reasoning from natural facts; I arrived at it because I am a person raised inside a moral tradition that treats human life as valuable and treats the taking of it as a category of wrong that warrants collective response. Other people, in other traditions, have held other views. Sociopaths, whose brains are wired differently, sometimes hold no view at all. The universality of my position among the people I know is not evidence that the position is post-ideological; it is evidence that the ideology is widely shared.
That a position is ideological does not make it wrong, and it does not make it undefendable. It just means that when one defends it, you are defending a specific normative commitment against other possible normative commitments, and the defense is going to require you to say what you think a person is and what you think a community is made up of and what you think we owe each other.
⁂
The claim to be post-ideological is a claim about oneself, but it is also, always, a claim about other people. When someone such as Amodei says ideology will not survive reality, he is not saying his own commitments will not survive reality. He is saying the commitments of people who disagree with his sense of what reality is will not survive the reality AI is about to reveal, and his own commitments will turn out to have been the plain description of what needed to be done.
It is almost never a claim of humility. It is a claim of transcendence — that the speaker has stepped outside the fog of partial interests and can see the shape of the actual situation while everyone else remains inside it. And the shape of the actual situation, remarkably often, turns out to correspond very closely to what the speaker already believed before he stepped outside the fog. So the Davos executive discovers that the necessary policy is a version of the policy he has been advocating for. The Silicon Valley engineer discovers that the necessary politics is a version of the technocracy he has been building companies inside of. The rationalist discovers that the necessary morality is a version of what he already thought was rational.
I do not think Amodei is doing this cynically. I think he genuinely believes that when he says ideology will not survive reality, he is describing an inevitable convergence rather than announcing his own preferences. And there is a weaker version of his claim that is defensible. Material pressure does move politics. Mass technological unemployment on the scale he predicts really would force some political response, and doing nothing really would become untenable. If Amodei were saying only that the scale of disruption will make a political settlement unavoidable, I would agree with him.
But that is not the claim he is making. He is not saying a settlement will be forced; he is saying this settlement will be forced — wage insurance, retraining programs, capital accounts, worker mobility support. These are defensible policies, and I would probably support most of them, but they are ideologically specific. What they come out of, is a particular tradition of technocratic liberalism that assumes market disruption should be met with redistributive smoothing rather than, say, structural constraints on the companies doing the disrupting, or public ownership of the AI models producing the displacement, or a moratorium on deployment until democratic institutions have caught up. Someone from any of those other traditions would look at the “bipartisan necessity” Amodei describes and find it recognizably not their view. Necessity can constrain the agenda without dictating the answer, and what Amodei has done is blur the two. He is describing his own view of the necessary and calling it the necessary.
The authoritarianism claim breaks even more cleanly, because Amodei has made the opposite claim inside the same essay. He warns that powerful AI in the hands of autocrats could amount to “an attempt to permanently steal their freedom and impose a totalitarian state from which they can’t escape,” and then, a few paragraphs later, offers the hope that AI will render autocracy unworkable the way industrialization rendered feudalism unworkable. Both cannot be true. Either AI is dissolving authoritarianism or it is entrenching it, and the current evidence — AI-enabled surveillance states, automated censorship, algorithmic control at a scale no previous technology allowed — runs strongly in the second direction. The reality will force convergence pose survives only if the speaker can keep updating what reality is forcing while never conceding he was guessing all along. This, you see, is the scientism tell. A genuine empirical claim about the direction of history could not hold both of these positions inside the same essay. A political preference dressed in the vocabulary of empirical claim can hold whatever it needs to hold in whatever paragraph it is in, because the vocabulary is decorative.
A version of this shows up in ordinary conversation too, from people who are not tech CEOs and have no policy program. It is the person who says I just want what works, or I don’t care about labels, I care about results, or both sides are broken and I refuse to pick. These sentences sound reasonable and sometimes are. What they smuggle in is a large assumption — that what works and results and broken are terms that describe themselves, that one can look at a policy and see whether it worked without first having decided what it was supposed to do and for whom.
There is no such view. A policy that reduces unemployment by driving down wages worked if your criterion was employment. It failed if the criterion was material conditions of workers. Whether it worked depends on what you were trying to do, and what you were trying to do depends on what you thought the situation required, and what you thought the situation required depends on the framework you brought to seeing it. That framework, you see, is your ideology.
The pretense is corrosive because it disables the argument the speaker most needs to have with himself. If one knows one is arguing from a specific position, one has to defend it against other positions, which means looking at what those other positions are asking to be seen. If one has convinced oneself that the argument comes from no position at all — that what is being described is simply the shape of the necessary — that work no longer has to be done. Disagreement, then, becomes evidence of the disagreer’s failure to see reality clearly.
The objection here is not particular to technocratic liberalism. It is an objection to a family of positions that share a structural error one might call materialist ethics — the belief that once the material situation has been correctly described, the ethical answer follows from it. Marxism does this when it argues that the mode of production determines the correct politics and that the emancipation of the working class is what history is empirically for. Right-libertarianism does it when it argues that the correct political arrangement follows from the natural facts of voluntary exchange under property rights, and that any deviation from that arrangement is coercion measurable against a physical baseline. Silicon Valley scientism does it when it argues that the correct politics follows from what advanced technology makes materially unavoidable. In all three cases, a moral vision is being smuggled into what presents itself as description. The Marxist has already decided that class emancipation is the point before the material analysis is run. The right-libertarian has already decided that individual freedom-from is the point before the natural rights derivation is run. Amodei has already decided that technocratic market-liberal smoothing is the point before the what reality forces analysis is run. What the description is doing, is concealing the choice, and the choice is what is actually up for argument. Materialism can constrain the space of possible answers, but the answer itself is a choice made inside that space by people with commitments. Anyone who tells you the material situation produced the answer has moved the choice offstage and is asking you not to notice.
This is not a small matter of intellectual manners. It is the operating system of the technocratic center for two generations, and it is what produced the populist moment we are living in. Daniel Bell declared the end of ideology in 1960. Tony Blair governed under the slogan what matters is what works, the notion being that rationality and pragmatism are the answer to ideological fanaticism. Bill Clinton did the American version. Anthony Giddens gave them the theory. A whole class of American commentators still writes in this register. The claim was always the same. We are not one faction among several. We are the ones who have set faction aside and are describing what the situation requires. What the situation turned out to require, in every case, was the specific preferences of the class making the claim: capital-friendly market centrism with a redistributive fringe, credentialed expertise given final say over questions the public was told were too technical to vote on, and a stern conviction that any resistance to this program was ignorance of the data. This class ran the West for forty years and told the people whose lives were being reorganized that they were describing the weather rather than governing them. When those people finally noticed, the response was not to concede that the framing had been a choice all along; the response was to double down and produce another paper showing that the numbers were actually strong.
The philosophical answer to what matters is what works is that a great many things work. Human beings have found many ways to live. The arrangements they have used for sharing resources, raising children, resolving disputes, and organizing production run to a substantial catalogue, and on the available evidence a fair number of them produce lives their inhabitants recognize as good. The plurality is not noise around a signal; it is the situation. This is Hume’s point at its widest — that no descriptive account of what human beings materially are can settle by itself what human beings ought to do together, because the ought is not contained in the is, and any argument that says otherwise has smuggled a value premise into what it calls description. The technocratic center smuggled a specific value premise: that market-friendly redistributive centrism was the arrangement the facts entailed. What Amodei is smuggling is a newer one: that whatever settlement follows from AI-driven displacement, on the terms his industry finds acceptable, is the arrangement reality is about to impose. Both smuggle. What actually follows from what works is the harder question of what it is that we are trying to make work, and for whom, and that is the ideological question the pose is designed to skip.
The empirical case against this move is now unusually clean. What the Washington consensus predicted, in explicit form, was that economic liberalization in China and Russia would produce political liberalization. Bill Clinton made the argument in 2000 at the podium, on behalf of permanent normal trade relations for China: “In the knowledge economy, economic innovation and political empowerment, whether anyone likes it or not, will inevitably go hand in hand.” He said liberty would spread by cell phone and cable modem. He said the internet would change China the way it had changed America. He said the more China liberalized its economy, the more fully it would liberate the potential of its people. This was not a fringe claim. It was the settled reading of the material situation across two presidential administrations, most of the American political class, and a wing of the academy that had inherited modernization theory from Lipset. The prediction was that once the material conditions were in place, the political outcome would follow because it had to. Twenty-five years later, China is a more capable authoritarian state than it was in 2000, wielding the tools of digital capitalism against its own people and against ours. Russia used its integration into global markets to fund the war it is currently prosecuting in Europe. Both regimes proved adept at running capitalist economies without conceding meaningful political freedom. The material conditions arrived. The political conditions did not follow. The theory was wrong at the scale of a quarter century and at the level of the country’s most consequential foreign-policy decisions, and its wrongness is the strongest available empirical demonstration that Hume was right — the is did not produce the ought, because the is never does. So Amodei is asking the country to make the same bet again, with the material lever moved from trade policy to AI capability, and with the same confidence that the political outcome he prefers is what the situation will impose.
A caveat about the lineage, and about the people inside it. Many of the thinkers who held the technocratic view held it in good faith. They believed that opening markets would raise living standards, that trade would knit interests together, that the material conditions they were assembling would produce the political outcomes they wanted, and they were sincere about the outcomes they wanted. The project they built made the world materially safer for global capital and politically more dangerous for democratic rights, and while many of them did not intend that outcome, the outcome is what history recorded, and many of them have not defended it since. The honest ones updated. Francis Fukuyama is the clearest case. His actual argument in the 1989 essay “The End of History?” was that liberal democracy had won the ideological argument, yes, but that the winning would produce a spiritual condition he described as a very sad time in which “the worldwide ideological struggle that called forth daring, courage, imagination, and idealism, will be replaced by economic calculation, the endless solving of technical problems, environmental concerns, and the satisfaction of sophisticated consumer demands.” When he expanded the argument into the 1992 book The End of History and the Last Man, the last man half of the title carried his warning about exactly the figure this piece is criticizing — the person who has stopped being able to imagine that anything else could be worth fighting for, and who mistakes that incapacity for the shape of reality. Fukuyama has since said in plain terms that the neoliberals went too far and that the United States now needs more social democracy, and, most tellingly, that he was never opposed to social democracy in the first place. He is not the target of this essay. The target is the pose that survives the update, worn now by a class of people who did not build the postwar consensus, did not watch it fail, and are refreshing its central move as if the last thirty years had not happened.
History did not end at home either. Less than three years after the 1989 essay, and within months of the book, Rodney King’s beating by four Los Angeles police officers was captured on video, the officers were acquitted on April 29, 1992, and Los Angeles burned for six days. The country that was said to have concluded its ideological argument was, in fact, still in the middle of an unresolved one about whether Black Americans were fully protected under the law it had already written, a question that was old when the end of history essay was written and is not settled now. The technocratic center could feel finished with argument because the argument had never been about the class writing the essays. The people the fight was actually about had never stopped making it, would not stop, and were told through those decades that their persistence in making it was itself evidence they had not caught up to reality. That is the pattern the pose imposes: the people who benefit from the current settlement pronounce the settlement to be reality, and the people it does not settle are told they have failed to grow up. It is the same move Amodei is making now, aimed at anyone whose commitments do not fit the settlement his industry is preparing to impose.
The populist right and the populist left both formed around the recognition that the technocratic center had chosen a specific arrangement and then insisted the arrangement was physics. They disagree about almost everything else. They agree about that. Amodei is not breaking from this lineage. What he is doing, is refreshing it for the AI era, from a seat with more concentrated power than any Third Way prime minister ever held. He is telling a country that has already once revolted against the what works frame that the frame is back and this time it comes with a computer. This is why the pose matters. The posture that produced the current politics through two generations of failure is the one Amodei has just picked up again, from a bigger seat and with a bigger claim.
⁂
The reason people reach for the post-ideological pose is that they have been convinced ideology is a slur. It isn’t. Ideology is what you get when you take human beings seriously as creatures who live together, who share resources, who make and break rules, who have to decide who counts as one of us and what we owe each other. These are questions that admit of answers, and the different answers arrived at by different traditions are what is meant when we talk about ideology. My answer to the question should murder be prohibited is that yes, murder should be prohibited, and it should be prohibited because human life is the kind of thing that makes claims on other human beings we are obligated to honor. That is a bedrock case, and I chose it because almost no one dissents from it, which is exactly the objection a skeptic could bring — that calling the prohibition of murder ideological stretches the word past the sense that matters. Fine. Take a harder case. What are the limits of private property? A libertarian says property rights are the near-absolute foundation of a free society and that redistribution is theft. A social democrat says property is a legal fiction that societies construct and reconstruct in service of human flourishing and that some forms of accumulation are themselves theft. Both answers are ideological. There is no neutral inspection of the situation that adjudicates between them. The word theft is doing different work in each sentence because each speaker has already answered the prior question of what property is for. That, you see, is what ideology is doing — constituting the framework in which the fact appears at all. The murder case and the property case are the same kind of thing. They differ only in how many people happen to agree.
The people I trust are the people who can name what they believe and why. The people I do not trust are the people who tell me they believe nothing in particular and are simply describing the situation. Ask them what they think the situation requires. Then ask what they thought the situation required last year, and the year before, and ten years ago. The answers will not be the same. What will be, is the tone.
That is the purest form of ideology — the form that has stopped noticing itself is a form at all. Amodei is not the first to strike this pose and he will not be the last. But he is a very clear case of it, and he is a smart enough man that if the mirror were held up carefully he would see what he is doing. Whether he would then stop doing it is a different question, and I suspect the answer is no, because the pose is doing too much work for him to give it up. What it is letting him do, is say his politics are physics. Once one has said that, one is hard to argue with, because there is nothing for the other person to argue with except the weather.
I would rather argue with someone who tells me what they believe. Give me the ideologue any day. At least the ideologue is in the argument.
Go Deeper into the Circus
Have a Good Life
A friend sent me a message on Signal last night and blocked me. I am writing about it not because my case is unusual but because it is not — because people have been losing friends and family over the past ten years in ways that follow this pattern almost exactly, and because I am one more person sitting with a phone that no longer has a name attached t…
The Enemy Is the Platform
It took me decades to see this. I came up inside libertarian and conservative circles, and I believed for most of my adult life that I was arguing within a movement that had ideas — flawed ideas, ideas I would eventually reject root and branch, but ideas. I extended the Republican Party the credit of assuming it believed things. That assumption survived…
I appreciate this so much! My brain has been canoodling these issues with a flashlight without being able to turn the bright overhead light on. You and the commenters have done that for me. Thank you. I archived it to read more deeply later.
In some ways, there always and still remains the question of, Yes, it works- but WHO does it work for??
The claim to post-ideology is just pretty new makeup for the politics of demonization. By collapsing everything into the post-ideology “singularity” that Dario et al espouse, all other ideologies become “irrelevant and unacceptable.” If the word ideology becomes “meaningless” then only people who agree with them can exist. It’s as naive and sophomoric as any other extremist belief. Wannabe autocrats always gonna autocrat.