The Technical Priesthood (Mumford Reading)
The technical priesthood is the third structural element of every megamachine, alongside the command center and the workforce. Mumford traced this element to the convergence that appeared in the river civilizations of Egypt and Mesopotamia: an astronomical priesthood whose ability to predict floods and seasons conferred an aura of supernatural knowledge upon the ruling apparatus. The priesthood's knowledge was genuine. Its predictions were accurate. And the accuracy of its predictions was precisely what made its authority so difficult to resist, because the authority derived not from coercion but from demonstrated competence. The AI priesthood — the machine learning engineers, systems architects, prompt designers, and alignment researchers who understand how large language models actually function — occupies the same structural position. Their knowledge is genuine; their competence is demonstrable; their outputs are impressive. The impressiveness of the outputs, not any threat of violence, secures the compliance of the population that uses the systems without understanding them.
In the AI Story
Mumford's framework adds a critical dimension to discussions of AI governance that emphasize the ethical intentions of AI developers. Edo Segal notes in The Orange Pill that people with deep understanding of AI systems bear an obligation to serve rather than exploit — that understanding confers not authority but stewardship. The observation is ethically admirable. But Mumford's historical framework suggests it underestimates the structural forces at work.
The Egyptian astronomers also believed they were stewards. The problem was not their intentions but their position: mediators between a system of extraordinary power and a population that could not independently verify the claims upon which that power rested. The priesthood's stewardship ethic could not survive the structural gap between its knowledge and the population's ability to evaluate the knowledge's application.
Mumford's analysis suggests three features that make the priesthood position structurally problematic regardless of individual character. First, the gap between specialized knowledge and general competence means that the population must accept the priesthood's framing of both problems and solutions, since it lacks the tools to formulate alternatives. Second, the priesthood's demonstrated competence in narrow domains (prediction, technical performance) becomes generalized into authority across domains it has not earned the right to address. Third, the priesthood's institutional position creates conflicts of interest — between the system's continued operation and the population's welfare — that no amount of personal integrity can fully neutralize.
These features apply to the contemporary AI priesthood with particular force. The gap between ML engineering expertise and general public competence is vast and growing. The demonstrated competence in narrow benchmarks becomes generalized into authority on questions (alignment, existential risk, economic displacement) that the technical expertise does not directly address. And the institutional position of the priesthood — employed by the same organizations whose systems they are supposed to evaluate — creates structural conflicts that stewardship ethics alone cannot resolve. Mumford's framework does not impugn the character of individual AI researchers; it identifies the structural position they occupy regardless of their intentions.
Origin
The concept developed across Mumford's work, receiving its fullest treatment in The Myth of the Machine (1967–1970). The Egyptian astronomical priesthood provided the paradigmatic case, but Mumford traced the structural pattern through medieval scholastic theologians, early modern natural philosophers, nineteenth-century engineers, and twentieth-century scientific managers.
Each iteration of the priesthood refined its mechanisms of legitimation while preserving the structural position. The AI priesthood's distinguishing feature is that it combines the legitimation mechanisms of previous forms — demonstrated prediction accuracy (Egyptian), esoteric knowledge (medieval), systematic method (modern scientific) — into a single institutional form.
Key Ideas
Structural position. The priesthood is defined by its mediating function between a system's power and a population's compliance, not by its members' intentions.
Genuine competence. Effective priesthoods do possess real expertise; the problem is not fraud but the structural gap their expertise creates.
Demonstration-based legitimacy. Authority derives from demonstrated success in specific domains, then generalizes to adjacent domains without independent validation.
Stewardship ethic insufficient. Individual commitment to serve the population cannot overcome the structural incentives of the priesthood position.
AI continuity. The AI priesthood occupies the same structural position as its historical predecessors, with refined but not transformed mechanisms.
Appears in the Orange Pill Cycle
Further reading
- Lewis Mumford, The Myth of the Machine, Vol. I (1967)
- Langdon Winner, The Whale and the Reactor (1986)
- Charles Lindblom, Inquiry and Change (1990) — on technical expertise and democratic governance
- Danielle Allen, 'How AI Fails Us' (2021) — contemporary extension of priesthood analysis to AI