The Industrial University is a Dead End
Why Efficiency Has Become Higher Education’s Greatest Vulnerability
Over the past several months, I have published a series of articles exploring what AI-resistant assessment might look like in an era of machine automation. Many solutions I have proposed, such as authentic assessments rooted in specific contexts or dialogic assessments built on sustained conversation, share a common thread. They demand something that our current educational infrastructure was explicitly designed to minimize: extended, individualized human contact between teachers and students.
The response to these proposals has been consistent and predictable. In conference discussions, comment sections, and email exchanges, educators have asked me some variation of the same question: “This sounds wonderful, but can I implement dialogic assessment when I teach three sections of 120 students each?”
My answer, increasingly, is that you cannot.
This is not a failure of imagination or pedagogical skill. The incompatibility runs deeper. What the emergence of AI has made acutely visible is that the strategies required to preserve meaningful education in an age of automated content production are fundamentally incompatible with the industrial model of education that has shaped our institutions for more than a century. This incompatibility is therefore not a pedagogical problem. It is an economic one. Understanding why requires examining how we arrived at this particular structural arrangement in the first place.
The Architecture of Scale: How Education Became Industrial
The modern university did not develop naturally into its current form. It was engineered according to specific principles borrowed from industrial manufacturing, principles that prioritized standardization and scalability above all else. The system we inhabit today is an artifact, constructed through deliberate choices made over decades. To understand its current fragility, we must first understand its construction. The following historical analysis tells the story of the US higher education system, but the same principles apply globally.
Prussian Foundations: The Standardization of Time
The American adoption of the Prussian educational model in the mid-19th century introduced what manufacturing would call “batch processing” to schooling. When Horace Mann, Secretary of the Massachusetts Board of Education, visited Prussia in 1843, he observed a system that had solved the logistical challenge of mass education through standardization. Students were sorted by age, moved through a uniform curriculum in annual cohorts, and assessed using common measures.
Prior to this intervention, education was erratic and individualized. The Prussian innovation made universal education administratively possible by treating students as interchangeable units moving through a standardized sequence. Every tenth-grader reads the same text at the same time, making it efficient to assess them all using the same techniques. The system assumed that if inputs (curriculum) and processes (time in class) were standardized, outputs could be standardized as well.
This structure enabled scale. Without uniform processing of students, the massive expansion of public education envisioned by democratic reformers would have been logistically impossible. But it also embedded a fundamental dependence: the entire system’s efficiency relied on the premise that standardized assessments could reliably measure learning.
The Cult of Efficiency: When Schools Became Factories
The decisive transformation occurred in the early 20th century during what historian Raymond Callahan documented as the “Social Efficiency Movement.” Influenced by Frederick Winslow Taylor’s principles of scientific management, which sought to optimize industrial production by breaking work into its smallest measurable components, educational administrators began viewing schools as factories and students as raw materials to be processed into finished products.
Taylorism demanded the separation of planning from execution. In factories, this meant managers designed workflows while laborers executed them. In schools, this produced a hierarchy where administrators set policy and teachers implemented standardized curricula, transforming faculty from autonomous intellectuals into managed laborers.
The efficiency movement required measurement. Schools were evaluated through surveys that assessed cost per student and facility utilization rather than intellectual depth or student growth. Standardized testing proliferated because it provided the quantifiable data needed to show efficiency to boards and taxpayers. This shift privileged forms of knowledge that were easily measurable—recall or procedural application—over those that were complex and idiosyncratic, such as critical thinking, synthesis, or creativity.
The system was optimized for assessability. The five-paragraph essay, the problem set, the research paper with standardized citation format—these artifacts were designed to be efficiently graded at scale. Their form mattered more than their substance, because form could be evaluated consistently across thousands of students by interchangeable graders.
The Currency of Scale: The Carnegie Unit
If the factory model provided the structure, the Carnegie Unit provided the currency. Introduced in 1906 by the Carnegie Foundation for the Advancement of Teaching as a mechanism to standardize faculty pensions, it defined academic credit based on “seat time”—approximately 120 hours of contact per year.
This invention accomplished three things, each with lasting consequences. First, it established time as a proxy for learning, decoupling credit from demonstrated competence and attaching it instead to exposure. You earn credit for being present, not for what you can do. Second, it made education interchangeable. Credits became commodities that could be banked, transferred, and accumulated across institutions. Third, it created the Student Credit Hour, the fundamental unit for state funding, faculty workload calculation, and tuition revenue. Universities could now treat education as inventory, calculating capacity by multiplying classroom seats by contact hours.
This infrastructure enabled the dramatic expansion of higher education in the 20th century. It also created a system where success was measured by throughput rather than transformation. When education is measured in seat time and demonstrated through standardized outputs, the system can scale indefinitely, as long as those outputs remain credible proxies for learning.
The Economic Trap: Commodifying the Inefficient
A key issue with the industrial model, which worsened in the late 20th century, was that efforts to make teaching more efficient risked undermining the quality of education. This phenomenon, known as Baumol’s Cost Disease, explains why higher education costs have risen inexorably even as class sizes have grown.
In manufacturing, technology makes labor cheaper. Automation increases output per worker, driving down costs per unit. In education, technology makes labor more expensive. As productivity rises across the economy, wages increase. Universities must compete for talent against the very industries achieving those productivity gains, but they cannot achieve similar gains themselves. A professor cannot meaningfully engage twice as many students without compromising the educational experience—or rather, they can, but only by accepting that compromise. To pay competitive salaries without raising tuition to the stratosphere, universities had to find a way to scale the unscalable.
The pressure was immense. State funding declined precipitously beginning in the 1980s. Federal support shifted from grants to loans. Institutions needed to generate more revenue while controlling costs. The solution, adopted broadly across American higher education, was to transform education into a standardized product that could be delivered at scale with minimal variable cost per unit.
This is where the industrial model’s careful engineering became essential. Because the system had already standardized inputs, processes, and outputs, it could increase throughput by simply increasing the ratio of students to faculty. The large lecture became the norm. Individualized feedback became cursory. The relationship between faculty and students became transactional rather than developmental. Teaching assistants and adjuncts, who effectively provide cheaper labor with less training, took on more of the actual instruction. And full-time faculty increasingly focused on research, which generates external funding and prestige.
The language of this transformation was revealing. In 1983, the report “A Nation at Risk” reframed education as a matter of economic competitiveness, positioning schools as suppliers of human capital to industry. Students became “clients.” Classes became “products.” Learning objectives became “deliverables.” Faculty were assessed by student evaluation surveys modeled on customer satisfaction metrics. The university became a business selling a service—the credential—rather than an institution facilitating a transformation.
This commodification was not simply ideological. It was an economic response to an impossible cost structure. If teaching cannot be made more efficient without losing quality, but costs must be controlled, the only option is to redefine quality in terms that permit efficiency. The industrial model accomplished this by shifting from education-as-formation to education-as-certification. What the university sold was not transformation but documentation. The credential became the product, and the credential was awarded based on the production of standardized outputs: essays, exams, or problem sets that could be evaluated quickly against rubrics.
The system worked because these outputs remained credible proxies. A student who could produce an acceptable research paper had presumably engaged in research. A student who could solve problem sets had presumably understood the material. The artifacts served as evidence of process. The grade certified not just the quality of the product but the occurrence of learning.
This was the compromise that allowed the industrial model to survive its economic contradictions. Scale was achieved by minimizing the expensive aspects of education—individualized attention, mentorship, human dialogue—and maximizing the scalable aspects: standardized content delivery and output-based assessment. As long as the outputs remained trustworthy indicators of process, the system could claim to be providing education rather than merely processing credentials.
The Dead End: When the Proxy Breaks
The emergence of generative AI therefore does not create a new scalability problem for higher education. It reveals an existing one. The industrial model’s economic compromise rested on one crucial assumption: that the production of certain artifacts required the cognitive processes we associate with learning. Remove that assumption, and the entire structure becomes indefensible.
AI has severed the link between product and process. It is now possible to create research papers with no need for research or synthesis. A problem set can be solved without understanding. The credential is accessible without educational transformation. The proxy is broken.
This is not a technological problem that better detection software can solve; it is a structural collapse. It exposes a hard truth: the industrial model never solved the problem of scaling education; it only solved the problem of scaling certification. As long as the credential was difficult to obtain, it retained value. The system worked because obtaining the credential required producing artifacts that demanded cognitive labor. AI has broken that chain. The industrial model promised to provide education efficiently at scale, but what it actually provided was only the appearance of education.
The implications cascade rapidly. Students are starting to think carefully about how to use their time and energy when they know AI can do the assignments. Faculty who recognize that they cannot reliably distinguish AI-generated work from student work face a choice between performative enforcement and pedagogical redesign. And once institutions acknowledge their credentials rely on outputs machines can generate, they must address a core question regarding their value proposition.
Efficiency, once the goal, is now the vulnerability.
The Human Alternative: What Machines Cannot Replicate
We do not lack evidence. We lack the economic will to act on it. The research on what actually produces educational outcomes is remarkably consistent, and it has been consistent for decades. Study after study identifies the same factor as the strongest predictor of student success, retention, and intellectual development: meaningful interaction between students and faculty. Not content delivery. Not institutional reputation. Not even curricular structure. Relationship.
Research on student-faculty interaction shows that these relationships mediate between engagement factors and educational outcomes. Students who develop meaningful connections with faculty members show higher levels of motivation, persistence, and academic achievement. The mechanisms are simple: faculty provide mentorship, model expert thinking, offer individualized feedback, and create accountability structures that support sustained effort.
Studies of High-Impact Practices, such as collaborative assignments, undergraduate research experiences, or capstone projects, reveal that effectiveness derives from structure rather than content. These practices work because they require extended interaction between students and faculty or among students in ways that cannot be scripted or standardized. They are, by design, unscalable.
Effective teaching is fundamentally dialogic. Learning happens through conversation, through the gradual construction of understanding via repeated interaction. These processes cannot be automated because they depend on responsiveness to individual learners at particular moments in their development.
The Stratification to Come
The crisis will not affect all institutions equally. Elite institutions with low student-faculty ratios, substantial endowments, and high tuition already operate closer to the relational model. They can credibly claim to provide formation rather than just certification. Their business model becomes more defensible as AI makes clear what distinguishes human-centered education from scaled credentialing.
Mass institutions, trapped by their cost structures, will face irresistible pressure to automate further. AI will be deployed for grading, tutoring, and content delivery. It will be deployed not as enhancement but as replacement. We are watching the emergence of a two-tier system that was always implicit but now becomes explicit: a boutique tier where the luxury of human mentorship remains available to the wealthy, and a mass tier where algorithms manage the instruction for everyone else. The industrial model does not disappear; it becomes automated.
This stratification is not inevitable, but preventing it requires recognizing that the problem is structural and economic rather than technological or pedagogical. The industrial model cannot provide value commensurate with its cost in a world where its primary outputs can be generated by machines. Institutions that continue operating on this model will find their market position increasingly untenable, not because educators have failed pedagogically, but because the economic foundation has dissolved.
The question is not whether the industrial model survives in its current form—it won’t. The question is whether we allow its collapse to proceed through market logic alone or whether we intervene to ensure that the transition serves democratic rather than merely economic ends.
What This Means for Practice
Meaningful change requires deliberate structural transformation, not individual heroics. When you design assignments that require human dialogue, when you reduce class sizes by offering fewer sections, or when you argue for hiring based on teaching capacity rather than research productivity, you are not simply making pedagogical choices. You are proposing a different business model for your institution.
The challenge is to make this case in terms that administrators and policymakers can understand. The argument cannot rest solely on pedagogical ideals. When arguing for smaller class sizes, do not rely primarily on “better learning” claims, which lose against “higher costs” in budget negotiations. Instead, demonstrate that the current model is economically unsustainable because it delivers a commodified product that machines now produce at near-zero cost.
Consider the calculation that students will increasingly make: Would you invest hundreds of thousands of dollars in an education built on an assessment system that AI has rendered obsolete? Would you pay premium prices for credentials based on outputs that machines generate at near-zero cost? The answer, for a growing number of students, will be no. The industrial model is not failing because it violates educational principles. It is failing because efficiency has become its primary vulnerability, and the very features designed to enable scale have created a system that machines can replicate without human participation.
This reframing is both sobering and clarifying. We are not debating the merits of different pedagogical approaches. We are witnessing the collapse of an economic model that has shaped education for a century. The institutions that survive this transition will be those that recognize this reality early enough to restructure around what AI cannot provide: sustained, unscalable, human interaction.
The industrial model is a dead end. We can stay on the bus as it drives off the cliff, automating a hollow credential until it becomes worthless. Or we can get out and start walking—slowly, inefficiently, and together—toward the only thing that ever worked.
The images in this article were generated with Nano Banana Pro.
P.S. I believe transparency builds the trust that AI detection systems fail to enforce. That’s why I’ve published an ethics and AI disclosure statement, which outlines how I integrate AI tools into my intellectual work.