An expert is not merely someone who has memorized data but someone who has internalized the structure of knowledge itself. This is why we call them PhDs—Doctors of Philosophy. Their expertise extends beyond isolated facts to the organizing principles that connect those facts, allowing them to wield knowledge in novel ways. While this definition is not airtight, it captures one key aspect of the priesthood that interests me: the structure of knowledge transmission.
The Priesthood of Knowledge
The concept of a priesthood in professional spheres comes from two sources: Astral Codex Ten’s recent discussion on expertise as a social function and Richard Posner’s 1988 article on Professionalisms.
Posner argues that professions—law, medicine, academia—function not merely as knowledge repositories but as cartels that control access to specialized knowledge, determining who is permitted to practice and what constitutes legitimate expertise. This gatekeeping is justified on the basis of quality control but often functions as self-preservation. Professionals claim authority by mastering a body of knowledge and its internal logic, which they enforce through credentials, ethical codes, and disciplinary norms.
Astral Codex Ten extends this idea, describing expertise as a priesthood—not in a pejorative sense, but as a structural necessity. Like religious orders, professions sustain their authority by maintaining interpretive dominance over their domain. They decide what counts as valid knowledge, who is qualified to interpret it, and how it should be disseminated. This ensures stability and continuity but also creates rigidity: professionals, like priests, can become blind to shifts in the structure of knowledge itself.
This priesthood model helps explain why professional authority is periodically challenged—when the structure of knowledge transmission shifts faster than the established class can adapt.
The Problem of Deciderization
This brings us to the problem of deciderization, a term borrowed from David Foster Wallace’s introduction to Best American Essays 2007. DFW describes “deciderization” as the process by which a system determines which stories, ideas, or facts matter and which do not—what is deemed important enough to be recorded, discussed, and remembered.
Deciderization is the hidden function of every knowledge system. It determines what makes it into textbooks, what gets cited in papers, and what is dismissed as irrelevant or trivial. The priesthood of knowledge plays a crucial role in this, curating and transmitting what is considered canonical. This is not just an academic exercise; it affects journalism (what counts as news vs. gossip), law (which precedents are binding vs. obsolete), and science (which data points are breakthroughs vs. noise).
Historically, professionals mediated the hierarchy of knowledge through their literacy in it. Other literate individuals could always learn the hierarchy, but the division of labor made professionalization valuable. However, professionalization also introduces fragility—jargon, restricted access, and institutional inertia. This means professionals struggle to adapt when the ground shifts beneath them.
The Fragmentation of Expertise
When knowledge expands faster than the priesthood can regulate, expertise fragments. This has two effects:
Challenge from below – Demand for expertise outstrips supply, leading to knowledge gaps filled by conspiracy theories, autodidacts, or parallel institutions. The Protestant Reformation, fueled by the printing press, is one historical example. More recently, the rise of blogs, online courses, and independent researchers has eroded centralized academic authority.
Fragmentation into sub-specialties – When a field becomes too broad, it splinters into specialized domains. The generalist disappears, replaced by specialists who defer to each other. Medicine provides a clear example: a small-town doctor no longer tries to cure everything but refers complex cases to specialists.
This fragmentation raises the question: what happens when AI accelerates the process further?
Why Doesn’t the Computer Age Get Its Own Era?
In considering professional revolutions, I proposed a four-stage model:
Oral Age (100,000–4000 BC)
Age of Writing (4000 BC–1453 AD)
Age of Print (1453–2024 AD)
Age of Oracles (2024–?)
But why not designate a separate "Computer Age," symbolized by Excel, the way printing gets its own epoch?
The key difference is that computers have accelerated the existing structures of writing and print rather than fundamentally altering them. Excel, for example, is “writing by other means.” It speeds up calculations, organizes data, and improves efficiency, but it does not revolutionize the fundamental act of recording, interpreting, or transmitting knowledge. The AI revolution, by contrast, represents a structural transformation—shifting authority away from those who master data structures and toward those who master queries and sanctions.
Rejoinder on Orality
A common claim is that we are entering a new age of orality, where writing becomes secondary to speech, conversation, and ephemeral digital media. But this misunderstands the historical pattern of technological change. The printing press did not displace the importance of communities of practice—it complemented them. In fact, print reinforced and elevated these communities, making expertise more geographically and institutionally concentrated rather than less. We see this in data on Nobel Prizes, where scientific excellence clusters in elite networks, and in the geography of total factor productivity (TFP) growth, where agglomeration effects drive innovation. Similarly, AI is not going to destroy writing; rather, it will increase the value of finding the best writing. As Tyler Cowen argues in In Praise of Commercial Culture, great increases in the volume and quality of low culture—in this case, an explosion of AI-generated text—expand the possibilities for high culture. The challenge is not the disappearance of expertise, but its restructuring: curation, evaluation, and authentication will matter more than ever.
AI and the Structure of Expertise
AI presents a fundamental shift in knowledge mediation. The role of the human expert will either be:
To validate the AI’s output (sanctioning authority)
To generate the AI’s input (structuring the problem)
Previously, humans rendered data into meaningful conclusions—turning is into ought, interpreting professional standards, and refining techniques through experience. Now, AI holds the knowledge while professionals become intermediaries between AI and society. Consider X-rays: once the sole domain of radiologists, they are now analyzed by machine learning models that doctors consult before delivering diagnoses. The doctor still plays a role because trust and liability require a human presence, but the cognitive authority is shifting.
The Adaptability Problem
The long-term impact of AI depends on how easily it can be modified:
If AI is hard to modify – Once an AI model adopts a method, human fine-tuning will be constrained. Errors and biases in training data could persist. Can we design systems that make re-training easier?
If AI is easy to modify – AI could integrate new discoveries constantly, updating itself dynamically. This would create a self-correcting knowledge system, but it also raises concerns about stability and interpretability.
The reality is likely a mix: AI will be adaptable in some areas but rigid in others. The implications will differ by field.
The Future of Expertise: Query Masters, Not Knowledge Masters
We are moving from an era where expertise meant mastering the structure of knowledge to one where expertise means mastering the structure of queries. Professionals will no longer be sole authorities; they will become arbiters of AI-generated outputs.
In this world, expertise will not be about knowing but about framing the right questions and sanctioning the right answers. The value of professionals will depend on how easily their roles can be bypassed by mass adoption of AI.
Some fields will remain resistant—where trust, liability, or ethical considerations require human judgment. But in other fields, the gatekeeping role of experts will erode. Those who understand how to train, benchmark, and interrogate AI systems will become the new high priests of knowledge.
An expert is not merely someone who has memorized data but someone who has internalized the structure of knowledge itself. This is why we call them PhDs—Doctors of Philosophy. Their expertise extends beyond isolated facts to the organizing principles that connect those facts, allowing them to wield knowledge in novel ways. While this definition is not airtight, it captures one key aspect of the priesthood that interests me: the structure of knowledge transmission.
The Priesthood of Knowledge
The concept of a priesthood in professional spheres comes from two sources: Astral Codex Ten’s recent discussion on expertise as a social function and Richard Posner’s 1988 article on Professionalisms.
Posner argues that professions—law, medicine, academia—function not merely as knowledge repositories but as cartels that control access to specialized knowledge, determining who is permitted to practice and what constitutes legitimate expertise. This gatekeeping is justified on the basis of quality control but often functions as self-preservation. Professionals claim authority by mastering a body of knowledge and its internal logic, which they enforce through credentials, ethical codes, and disciplinary norms.
Astral Codex Ten extends this idea, describing expertise as a priesthood—not in a pejorative sense, but as a structural necessity. Like religious orders, professions sustain their authority by maintaining interpretive dominance over their domain. They decide what counts as valid knowledge, who is qualified to interpret it, and how it should be disseminated. This ensures stability and continuity but also creates rigidity: professionals, like priests, can become blind to shifts in the structure of knowledge itself.
This priesthood model helps explain why professional authority is periodically challenged—when the structure of knowledge transmission shifts faster than the established class can adapt.
The Problem of Deciderization
This brings us to the problem of deciderization, a term borrowed from David Foster Wallace’s introduction to Best American Essays 2007. DFW describes “deciderization” as the process by which a system determines which stories, ideas, or facts matter and which do not—what is deemed important enough to be recorded, discussed, and remembered.
Deciderization is the hidden function of every knowledge system. It determines what makes it into textbooks, what gets cited in papers, and what is dismissed as irrelevant or trivial. The priesthood of knowledge plays a crucial role in this, curating and transmitting what is considered canonical. This is not just an academic exercise; it affects journalism (what counts as news vs. gossip), law (which precedents are binding vs. obsolete), and science (which data points are breakthroughs vs. noise).
Historically, professionals mediated the hierarchy of knowledge through their literacy in it. Other literate individuals could always learn the hierarchy, but the division of labor made professionalization valuable. However, professionalization also introduces fragility—jargon, restricted access, and institutional inertia. This means professionals struggle to adapt when the ground shifts beneath them.
The Fragmentation of Expertise
When knowledge expands faster than the priesthood can regulate, expertise fragments. This has two effects:
This fragmentation raises the question: what happens when AI accelerates the process further?
Why Doesn’t the Computer Age Get Its Own Era?
In considering professional revolutions, I proposed a four-stage model:
But why not designate a separate "Computer Age," symbolized by Excel, the way printing gets its own epoch?
The key difference is that computers have accelerated the existing structures of writing and print rather than fundamentally altering them. Excel, for example, is “writing by other means.” It speeds up calculations, organizes data, and improves efficiency, but it does not revolutionize the fundamental act of recording, interpreting, or transmitting knowledge. The AI revolution, by contrast, represents a structural transformation—shifting authority away from those who master data structures and toward those who master queries and sanctions.
Rejoinder on Orality
A common claim is that we are entering a new age of orality, where writing becomes secondary to speech, conversation, and ephemeral digital media. But this misunderstands the historical pattern of technological change. The printing press did not displace the importance of communities of practice—it complemented them. In fact, print reinforced and elevated these communities, making expertise more geographically and institutionally concentrated rather than less. We see this in data on Nobel Prizes, where scientific excellence clusters in elite networks, and in the geography of total factor productivity (TFP) growth, where agglomeration effects drive innovation. Similarly, AI is not going to destroy writing; rather, it will increase the value of finding the best writing. As Tyler Cowen argues in In Praise of Commercial Culture, great increases in the volume and quality of low culture—in this case, an explosion of AI-generated text—expand the possibilities for high culture. The challenge is not the disappearance of expertise, but its restructuring: curation, evaluation, and authentication will matter more than ever.
AI and the Structure of Expertise
AI presents a fundamental shift in knowledge mediation. The role of the human expert will either be:
Previously, humans rendered data into meaningful conclusions—turning is into ought, interpreting professional standards, and refining techniques through experience. Now, AI holds the knowledge while professionals become intermediaries between AI and society. Consider X-rays: once the sole domain of radiologists, they are now analyzed by machine learning models that doctors consult before delivering diagnoses. The doctor still plays a role because trust and liability require a human presence, but the cognitive authority is shifting.
The Adaptability Problem
The long-term impact of AI depends on how easily it can be modified:
The reality is likely a mix: AI will be adaptable in some areas but rigid in others. The implications will differ by field.
The Future of Expertise: Query Masters, Not Knowledge Masters
We are moving from an era where expertise meant mastering the structure of knowledge to one where expertise means mastering the structure of queries. Professionals will no longer be sole authorities; they will become arbiters of AI-generated outputs.
In this world, expertise will not be about knowing but about framing the right questions and sanctioning the right answers. The value of professionals will depend on how easily their roles can be bypassed by mass adoption of AI.
Some fields will remain resistant—where trust, liability, or ethical considerations require human judgment. But in other fields, the gatekeeping role of experts will erode. Those who understand how to train, benchmark, and interrogate AI systems will become the new high priests of knowledge.