Course design in the age of AI

TEXT START: I develop a model of learning-by-doing and course design, and use it to study the impacts of artificial intelligence (AI).

The Dissection

The paper models AI as a delegation option inside education. Its real subject is how instructors can redesign tasks to force humans to keep practicing when outsourcing becomes effortless. It identifies a genuine early-stage failure mode: AI can hollow out the learning process before institutions adapt, while improvements in AI widen the gap between students who can exploit it and those who cannot.

But this is not a full theory of the age of AI. It is a local model of skill formation that leaves the labor market, ownership, bargaining power, and the economic necessity of human labor outside the frame.

The Core Fallacy

Under the Discontinuity Thesis, the core error is treating human skill development as the strategic endpoint. The model asks how to produce more capable humans while holding fixed the assumption that the economy will still need to purchase their capabilities.

That assumption is the entire battlefield. Under P1–P3, better-trained humans may remain relatively superior to worse-trained humans, yet still become economically unnecessary once AI achieves durable cost and performance dominance. The paper confuses the ability to create human skill with the economy’s willingness or need to pay for it.

Its internal result may be valid. Its strategic conclusion is incomplete: curriculum redesign is a lag defense, not a reversal of labor obsolescence.

Hidden Assumptions

• Human skill remains scarce, valuable, and economically necessary after AI improves.
• Teachers and institutions can reliably force effort and restrict delegation.
• Students remain myopic, rather than learning to use AI for verification, orchestration, or accelerated practice.
• Delegation produces no transferable skill, while AI-assisted learning does not itself become a substitute for human mastery.
• Course redesign can scale despite institutional weakness and competition from easier AI-mediated alternatives.
• Skill gaps, rather than ownership of AI capital, are the decisive source of economic status.
• AI primarily complements effort instead of replacing the tasks for which the effort was being purchased.
• Credentials and curricula retain labor-market authority after productive participation collapses.

Social Function

This is partial truth packaged as transition management, with an ideological-anesthetic edge.

The truthful part is that unstructured AI use can destroy learning-by-doing and amplify inequality. The managerial function is to relocate a civilizational problem into syllabus engineering, allowing education systems to claim that better task design can preserve human capability.

The anesthetic is the continued assumption that capability preservation equals economic survival. It does not. A better-trained worker is still a worker, and the thesis concerns the disappearance of the need for most workers.

The Verdict

This paper performs a competent autopsy on one organ of the dying system: AI can degrade the human-capital pipeline before it destroys the labor market itself. It does not diagnose the corpse.

Course redesign may produce more capable Servitors and delay skill atrophy. It cannot solve coordination impossibility, restore mass productive participation, or give students control over AI capital. Its most important implication is not educational rescue but stratification: high-skill users accelerate, low-skill users lose practice, and the gap hardens into a selection mechanism.

The paper describes how to postpone human obsolescence. It provides no mechanism for preventing it.