Hierarchist Response to AI
The hierarchist response to AI is the cultural position most visible in the professional and regulatory response to the technology. It interprets AI through the lens of order: the maintenance of standards, the preservation of credentialing systems, the orderly integration of new capability into existing institutional structures. Hierarchists are sensitive to the risks of uncontrolled deployment — AI-generated misinformation, the erosion of professional competence, the breakdown of established quality mechanisms. Their preferred remedies are institutional: licensing regimes, professional standards bodies, regulatory oversight, expert certification. Their characteristic blind spot is the capture of the institutions they trust, and the possibility that the disorder they fear is more productive than the order they defend.
The Substrate of Control — Contrarian ^ Opus
There is a parallel reading that begins not with cultural positions but with material dependencies. The hierarchist response to AI isn't merely a cultural preference for order — it's the predictable behavior of institutions whose survival depends on maintaining information asymmetry. Professional credentialing systems evolved when knowledge was scarce and verification costly. They solved a real coordination problem: how to signal competence when most people couldn't evaluate it directly. But this solution created its own political economy — entire sectors organized around the artificial scarcity of certified judgment.
The hierarchist institutions now proposing to govern AI are precisely those whose monopoly on legitimation is threatened by it. The medical boards watching AI diagnostic tools outperform their members, the law societies seeing contract generation automated, the academic departments whose peer review AI can simulate — these aren't neutral arbiters of public safety but interested parties in a struggle over economic rents. Their proposals for licensing and oversight reproduce the same gatekeeping logic that protected their predecessors from the printing press, the internet, and every other technology that democratized access to formerly restricted knowledge. What they frame as protecting the public from AI's risks is indistinguishable from protecting their position as intermediaries. The compute thresholds and mandatory evaluations they propose will, in practice, create regulatory moats that only well-capitalized firms can cross — replacing one form of professional gatekeeping with another, this time with the gatekeepers drawn from the very firms they purport to regulate.
In the AI Story
The hierarchist reading sees the professional as the carrier of quality. Decades of training, credentialing, peer review, and institutional socialization produce the judgment that distinguishes reliable work from plausible output. AI threatens this system at the root, because it produces plausible output without the judgment, and because it lowers the barrier between the credentialed practitioner and the informed amateur. The hierarchist is right that something is at stake here — the very architecture of how societies distinguish the expert from the charlatan — but the response often reduces to defending the gate rather than asking what the gate was for.
The Orange Pill's observation that AI removes the translation cost between intention and artifact reads as threatening to the hierarchist for precisely this reason. If the non-credentialed developer can build what previously required the credentialed one, what does the credential mean? One answer — the hierarchist answer — is that the credential should be strengthened, its boundaries policed more carefully, its authority defended against the leveling tendency of the tool. Another answer — the one Wildavsky would have offered — is that the credential's meaning has changed, and the institution's task is to find the new boundary between judgment (still scarce) and execution (now abundant), rather than to defend the old one.
The contemporary AI safety establishment is substantially hierarchist in its cultural logic. International coordination, compute thresholds, mandatory evaluations, licensing of frontier models — these are hierarchist proposals, structured around the idea that a competent expert body can identify the relevant risks and impose appropriate controls. The egalitarian critique of this approach — that the expert body will be captured by the firms it regulates — is the characteristic friction between the two cultural positions.
Wildavsky's relationship to the hierarchist position was complex. He respected institutions and believed pluralism required their presence, but he was acutely aware of their failure modes: ossification, capture, the production of rules that protected the rulemakers rather than the public. His preferred institutions were those with strong feedback mechanisms — those that could fail visibly and correct quickly. An AI governance regime that meets this standard remains hypothetical; most current proposals do not.
Origin
The hierarchist position is the cultural home of bureaucratic institutions, professional associations, and regulatory agencies. Applied to technology, it produces the credentialing and oversight apparatus that governs pharmaceuticals, aviation, and financial services.
The hierarchist reading of AI is visible in the proposals of institutions like the EU's AI Office, the UK AI Safety Institute, and the internal governance teams at frontier AI labs. Its strongest intellectual defenders argue that AI is too powerful to leave to market dynamics or individual judgment, and that only coordinated institutional response is adequate to the scale of the technology.
Key Ideas
Order is the primary value. The risk is disorder — the breakdown of the systems that distinguish reliable from unreliable, expert from amateur.
Credentials carry judgment. Professional training produces not just skill but the capacity to evaluate, which AI threatens to decouple.
Expert institutions as remedy. Licensing, certification, and regulatory oversight are the characteristic responses.
Coordination over competition. International coordination and standardization are preferred to market-driven discovery.
The failure mode is capture. Hierarchist institutions ossify and are captured by the firms they regulate — the chronic vulnerability of the position.
Debates & Critiques
The most consequential internal debate among hierarchists concerns which institutions should govern AI — national regulators, international bodies, professional associations, or the AI firms themselves through internal governance. Each answer produces a different political program and different vulnerabilities to capture.
Appears in the Orange Pill Cycle
The Necessity of Friction — Arbitrator ^ Opus
The tension between these views depends critically on which timeframe we examine. In the immediate term — say, the next five years — the contrarian reading dominates (75/25). Existing hierarchist institutions are indeed defending territory more than public interest, and their proposed governance structures do risk regulatory capture by incumbent AI firms. The EU's AI Act and similar frameworks read more as attempts to preserve institutional relevance than as genuine responses to novel risks.
But zoom out to the decade-scale and the hierarchist concern becomes more legitimate (60/40 in their favor). Some form of quality control and verification will be necessary as AI systems become more powerful and pervasive. The question isn't whether we need institutions to manage AI's integration into critical systems — we do — but rather what form those institutions should take. The hierarchist error is assuming existing professional bodies can simply extend their mandate; the contrarian error is assuming no institutional response is needed at all.
The synthesis emerges when we reframe the question from "who should control AI" to "what friction should AI systems encounter?" Both views agree that pure unconstrained deployment is dangerous, and both recognize that current institutions are inadequate. The productive path forward likely involves new institutional forms that combine the hierarchist instinct for standards with feedback mechanisms that prevent capture — perhaps through transparent benchmarking, rotating governance, or novel accountability structures that didn't exist in the pre-AI world. The real work isn't defending old gates or tearing them down, but designing new filters that can distinguish genuine risk from professional protectionism.
Further reading
- Stuart Russell, Human Compatible (Viking, 2019)
- Nick Bostrom, Superintelligence (Oxford University Press, 2014)
- Gillian Hadfield, Rules for a Flat World (Oxford University Press, 2017)
- European Union, AI Act (Regulation EU 2024/1689)