Protecting Engineers' Skills in the AI Era

TEXT START: A little over a decade ago, I led the controls design for a first-of-its-kind full digital-control system for a U.S. nuclear plant.

The Dissection

The text is really arguing that apprenticeship is infrastructure, not overhead. Its “manual gate” preserves human competence by forcing engineers to perform formative work before AI is allowed to assist.

That is a valid local remedy for skill decay in safety-critical systems. The article correctly identifies the broken mechanism: AI can eliminate the junior work through which senior judgment was historically produced. It also admits the political weakness of its proposal: quarterly-governed firms have little incentive to fund capabilities whose payoff arrives years later.

But the text quietly expands a niche safety protocol into a defense of the engineering career ladder. That leap is where the argument fails.

The Core Fallacy

The article confuses preserving a skill with preserving economic necessity.

A manual gate can keep a human capable of debugging. It cannot make that human the cheapest, fastest, or most scalable producer of debugging work. Under P1, AI achieves durable superiority across cognitive work. Under P2, institutions cannot preserve large human-only economic domains through goodwill or procedure. Under P3, the majority lose access to economically necessary labor.

Manual gates are therefore not a reversal of automation. They are contingency engineering inside an automated system. They may preserve a small number of human operators where failure costs justify the premium. They do not restore the mass apprenticeship ladder.

The aviation analogy is also structurally limited. Aircraft are regulated, physically constrained, and catastrophically expensive to operate without qualified humans. Most software and engineering work does not carry those same mandatory human-retention requirements. A firm may preserve debugging practice in a critical module while automating nearly every other entry-level task. The result is not a healthy career pipeline. It is a narrow reserve of trained Servitors maintained around AI capital.

The article’s “bad day” premise may be true, but even then the implication is brutal: one current human may supervise many automated systems. Human competence becomes an insurance asset, not a mass source of employment.

Hidden Assumptions

• AI will remain a tool that occasionally fails, rather than a system capable of generating, testing, simulating, and improving its own fallback procedures.
• Periodic manual practice is sufficient to preserve expert judgment under rare and novel failures.
• Firms, founders, or regulators will pay indefinitely for human capability that is mostly unused.
• Preserving junior work will lead to enough future senior positions to justify the pipeline.
• Safety-critical engineering economics can be generalized to ordinary engineering and software work.
• Human expertise has economic value merely because it has operational value. Those are not the same thing.
• Apprenticeship must remain human-to-human rather than being partly or largely reproduced through AI simulation, synthetic failures, and automated evaluation.
• Keeping a small cadre skilled is equivalent to preserving broad productive participation. It is not.

Social Function

The text is a partial truth serving as transition management and elite self-exoneration.

It gives institutions a respectable way to preserve a thin layer of human expertise without confronting the larger consequence: AI may destroy the entry-level ladder that creates that expertise. “Deliberate inefficiency” is genuine insurance in nuclear power or aviation. Applied across the labor market, it becomes a comforting label for a costly exception that most firms will not fund.

The article is not crude propaganda. Its diagnosis of skill atrophy is real. Its social function is subtler: make the transition appear governable through better workflow design while leaving the ownership structure and labor displacement untouched.

The Verdict

This is a sound cockpit procedure mistaken for a rescue plan for the labor system.

Manual gates can preserve critical human skills in regulated enclaves. They can create Servitor, transition-intermediation, and verification niches. They cannot defeat P1, P2, or P3. They do not reopen the junior career ladder, restore mass productive participation, or repair the wage-to-consumption circuit.

The human the article preserves is no longer the ordinary worker trained through production. He is a maintained insurance component around AI capital. That is not the survival of post-WWII capitalism. It is a more orderly form of its obsolescence.