In 1975, the Trilateral Commission — founded by David Rockefeller, directed by Zbigniew Brzezinski — published a report on what it called The Crisis of Democracy. The crisis was not that democracy was failing. It was working too well.

Samuel Huntington, the Harvard political scientist who wrote the American chapter, diagnosed the problem: "Some of the problems of governance in the United States today stem from an excess of democracy." The "marginal social groups" he named as examples were Black Americans. Their new participation in politics was, in his framing, part of the disease. The remedy:

[T]he effective operation of a democratic political system usually requires some measure of apathy and noninvolvement on the part of some individuals and groups.

He closed by quoting John Adams — "Democracy never lasts long. It soon wastes, exhausts, and murders itself" — and concluded: "There are also potentially desirable limits to the indefinite extension of political democracy." Five members of the Trilateral Commission entered the Carter administration the following year.

The report's introduction cited Walter Lippmann by name. Lippmann had declared the founding premise of self-governance dead fifty years earlier in Public Opinion: "It is no longer possible, for example, to believe in the original dogma of democracy." His alternative was a "specialized class" that "acts upon information that is not common property, in situations that the public at large does not conceive." In The Phantom Public, he sharpened the point: "The public must be put in its place... free of the trampling and the roar of a bewildered herd."

Edward Bernays cited Lippmann by name and got to work. The opening of Propaganda (1928): "The conscious and intelligent manipulation of the organized habits and opinions of the masses is an important element in democratic society. Those who manipulate this unseen mechanism of society constitute an invisible government which is the true ruling power of our country." He meant it as a compliment.

Tom Nichols, writing in The Federalist ninety years later, mourned a "Google-fueled, Wikipedia-based, blog-sodden collapse of any division between professionals and laymen." Citizens' demand for equal voice was "a manic reinterpretation of 'democracy' in which everyone must have their say." In a 2021 podcast, he said what Lippmann would not: "Let's just call them what they are — paternalistic solutions."

A hundred years of argument — Lippmann through Bernays through the Trilateral Commission to Nichols — all pointed one direction: the public cannot be trusted, so someone else must govern.

The Founders had a different problem in mind.

Madison, 1822: "A popular Government, without popular information, or the means of acquiring it, is but a Prologue to a Farce or a Tragedy." Jefferson, 1779: "Those entrusted with power have, in time, and by slow operations, perverted it into tyranny." Jefferson again, First Inaugural: "Sometimes it is said that man can not be trusted with the government of himself. Can he, then, be trusted with the government of others?"

Madison warned in Federalist 39 that republican government must be "derived from the great body of the society, not from an inconsiderable proportion, or a favored class of it." Lippmann's "specialized class," acting on "information that is not common property," is exactly the favored class Madison ruled out. Jefferson's 1779 bill assumed citizens could "know ambition under every disguise it may assume; and knowing it, to defeat its views." Lippmann's 1925 book argued they could not. The apathy doctrine reversed the founding design.

For most of the twentieth century, the apathy doctrine was arguably logical. A citizen in 1922 could not independently evaluate foreign policy, monetary policy, or regulatory capture. Expertise was expensive, and analysis was gatekept. A citizen in 1975 could not read the Trilateral Commission's report, cross-reference it against Lippmann, verify Huntington's citations, and publish the synthesis — not without institutional resources ordinary people did not have. The information asymmetry was real. It was the load-bearing wall of the whole arrangement.

But in the AI age, the wall is coming down.

A Small Wars Journal assessment put a number on it: small OSINT cells now generate 70-90 percent of the analytic value of classified collection at roughly two percent of the cost.

A single researcher with AI tools can process 2,000 pages in one analysis run, work that took a team of research assistants weeks. ProPublica used an LLM to analyze 3,500 NSF grant abstracts. The New York Times transcribed 500 hours of leaked video into five million words of searchable text — an investigation that would have missed its deadline without AI. Heritage Foundation filed over 50,000 FOIA requests in two years. Brown University's AISLE portal tracks over 1,000 AI-related bills across all fifty states. I built a setup that tracks 913,000 nonprofits and 1.3 million federal grants. That used to take teams at the GAO.

The obvious objection: AI hallucinates, amateurs make mistakes, speed without institutional checks produces noise. Nichols anticipated this a decade ago — the democratization of knowledge tools would only deepen the Dunning-Kruger spiral.

Fair. But the baseline is the institutional track record. The intelligence community's Iraq WMD assessment was a "major intelligence failure" that cost over 4,000 lives. Credit rating agencies enabled $22 trillion in losses in 2008. Twenty-seven scientists published a Lancet letter calling the lab-leak hypothesis a "conspiracy theory," organized by a researcher with an undisclosed conflict of interest, suppressing legitimate inquiry for eighteen months.

By comparison, the canonical AI hallucination case — Mata v. Avianca — was caught and sanctioned within months. A 2026 PLOS ONE study found that crowdsourced fact-checks and expert fact-checks were equally effective at reducing belief in misinformation — but the crowdsourced checks arrived faster.

The gatekeeping argument is also circular. Legitimate knowledge is knowledge produced by credentialed institutions. Knowledge produced outside them is illegitimate. When an independent researcher produces work that matches institutional quality, the objection shifts from "amateurs can't do this" to "amateurs shouldn't do this." That's Huntington's argument with sour grapes.

The productivity multiplier isn't going to transform the nonprofit sector. Only 32% of foundation leaders say they provide general operating support to most grantees — the rest lock recipients into specific spending categories, where efficiency gains just vanish. TechSoup's 2025 benchmark found that only 7% of nonprofits have successfully adopted AI, while 76% lack even a strategy for it. Nearly a quarter of foundations refuse AI-generated grant applications.

The reason is simple: AI’s productivity multiplier threatens the information asymmetry these institutions depend on. AI doesn't speed up the bureaucracy; it routes around bureaucracy. The growing class of independent researchers — OSINT analysts, data journalists, grant investigators, policy trackers — are already informing policy faster than the expertise class that is telling them to sit down.

Jefferson's answer to an uninformed citizenry was not to remove them from governance: "to inform their discretion by education. This is the true corrective of abuses of constitutional power." Madison's worry was not democratic excess. It was concentrated ignorance: "Knowledge will forever govern ignorance."

For a hundred years, the apathy doctrine won because it claimed to have the facts. According to them, the specialized class was necessary because citizens could not process the complexity of modern governance.

AI changed the facts, by making every citizen capable of verifying what the experts claim. The cost of analysis has now dropped by orders of magnitude.

The credentialed class has already anticipated this and dismissed this argument. "Universal literacy was supposed to educate the common man to control his environment," Bernays wrote. "[I]nstead of a mind, universal literacy has given him rubber stamps." But by that logic, the printing press was a failed experiment too. It handed the masses text without understanding—yet it also broke the Church's monopoly on scripture, and what followed was the Reformation.

The Founders built a system on the consent of the governed. For a century, a competing tradition argued that consent was a fiction best left unexamined. A twenty-dollar AI subscription has now made the specialized class optional.

If you enjoyed this, please pre-order the upcoming book I co-authored with @JoshuaLisec, Unelected, available on Amazon, Barnes and Noble, and Books-a-Million.