The Internship Apocalypse Is Here | Washington Monthly

TEXT START: Last recruiting cycle, Kevin Li, a rising senior at Duke University, applied to somewhere between 150 and 200 internships.

THE DISSECTION

The article documents an early fracture in the labor-market training pipeline. AI has not merely reduced interns’ workload; it has exposed that much of the internship existed as cheap labor, make-work, and supervised screening. One intern completes months of estimated work in two weeks, then deliberately throttles output because the institution still pays for occupied time rather than delivered value.

The remaining function is selection. Internships survive where employers use them to identify future hires, not because interns are economically necessary. That distinction is the article’s most important fact: the internship is being stripped of production and retained as a filtering ritual for scarce human positions. The result is fewer slots, more applicants, higher status concentration, and a sharper divide between candidates admitted into the pipeline and everyone else.

THE CORE FALLACY

The article treats the destruction of entry-level work as a redesign problem. Its proposed answer—passion projects, open-ended assignments, and better talent identification—confuses a temporary allocation problem with a structural one.

If AI can perform the routine work, the firm does not need to manufacture equivalent human work to preserve mass entry. It can simply hire fewer people and use the internship as a more competitive audition. “Meaningful” projects do not restore broad labor demand; they become higher-resolution tests for a smaller number of candidates. The article mistakes the survival of recruiting for the survival of productive participation.

Its opening reassurance that analyst, developer, and paralegal headcounts have grown since 2022 is also weak evidence against the Discontinuity Thesis. It measures the lag phase, while the internship market reveals the mechanism already operating at the margin: one worker, armed with AI, absorbs the output previously distributed across several junior workers. Headcount growth during expansion is not proof of durable human necessity. It is compatible with a system approaching the point where productivity gains outrun labor demand.

The article also underestimates coordination impossibility. It assumes firms can preserve a stable human-only developmental layer through better managerial design. But every firm faces competitive pressure to automate, compress onboarding, and select only the humans who add scarce judgment, trust, access, maintenance, or control. No collective preference for “meaningful work” can stabilize a mass internship system against that pressure.

HIDDEN ASSUMPTIONS

• That employers will continue to value broad human talent pipelines even when AI can perform most junior work.
• That internships can remain widespread as recruiting instruments despite becoming economically unnecessary.
• That open-ended projects will create enough real work to absorb the displaced population.
• That the knowledge needed to make interns highly useful will remain available through human onboarding rather than becoming another bottleneck or being encoded into AI systems.
• That rising conversion rates indicate opportunity rather than intensified selection among a shrinking pool.
• That a degree, major, GPA, or prior internship can retain meaningful signaling power when AI makes credentialed applicants increasingly interchangeable.
• That the transition will be judged by whether internships become more intellectually interesting, rather than by how many people lose access to the economic ladder altogether.
• That the labor market can preserve a mass rite of passage after the mass need for junior labor has been removed.

SOCIAL FUNCTION

Classification: partial truth, transition management, and ideological anesthetic.

The article is not pure copium. It accurately captures the early kill mechanism: routine cognitive work disappears first, output is concentrated into fewer workers, and firms retain internships primarily as elite recruitment filters. It also identifies the psychological farce produced when an employee can finish the assigned work almost instantly but must simulate effort for the remainder of the day.

But it softens the systemic implication by presenting contraction as modernization. “Passion projects” and more meaningful summers are the respectable language of a labor market that is closing its lower rungs. The students are told the old ritual was pointless just as access to the replacement pathway becomes scarcer. That reframes exclusion as liberation and converts a distributional catastrophe into a design critique.

The article’s hidden ideological service is to preserve faith in the professional pipeline: internships may shrink, but the worthy will supposedly receive better opportunities and stronger conversion odds. That is transition management for the credentialed class. It describes the gate narrowing, then calls the remaining gate more valuable.

THE VERDICT

The internship apocalypse is not an isolated youth-market anomaly. It is an early-warning indicator of productive participation collapse. AI is eliminating the cheap, low-stakes work through which firms historically trained and selected new workers. What remains is not a healthier ladder but a narrower audition controlled by those who own or command the productive systems.

The article sees the first corpse but misidentifies the cause of death. The internship is not primarily being rescued from meaningless work; it is being converted from a labor institution into a rationing mechanism. As AI capability spreads, the mass internship dies first, followed by the assumption that every educated young person can earn a place in the white-collar economy through effort, credentials, and persistence.