Is learning on the job making a comeback in tech? - Spiceworks

TEXT START: As AI changes the skills companies need, employers may be finding that hiring their way out of the skills gap isn’t enough.

The Dissection

This article documents a transition-stage labor tactic, not a revival of secure tech careers. Firms retrain existing workers because AI deployment creates immediate workflow gaps, institutional knowledge remains human-held, and replacement is still costly and slow.

The IBM–OpenAI partnership is presented as worker opportunity, but it primarily demonstrates concentration. The durable assets are model control, compute, proprietary data, customer access, and automated workflows—not individual certificates.

The Core Fallacy

The article confuses temporary complementarity with durable human necessity. Retraining can raise a worker’s near-term productivity while making the worker easier to replace later. The same AI that teaches a workflow can encode, standardize, monitor, and eventually execute it.

Low reported layoffs prove only that displacement is lagged. Under the Discontinuity Thesis, P1 advances as cognitive work becomes cheaper to automate; P2 prevents institutions from preserving human-only domains at scale; P3 follows as workers lose access to economically necessary labor. A trained surplus is still a surplus.

Hidden Assumptions

• Retraining produces upward mobility rather than redeployment, workload intensification, or delayed displacement.
• Employers will keep investing in workers after AI can transfer and perform their knowledge.
• Prompt engineering and basic AI literacy will remain scarce instead of becoming commodity interface skills.
• Productivity gains will expand jobs rather than consolidate them into fewer high-leverage roles.
• Adaptability makes workers indispensable rather than merely useful until a better agent arrives.
• Microlearning and certifications create durable market power despite the article admitting that credentials rapidly decay.
• The absence of layoffs captures displacement, ignoring hiring freezes, attrition, nonreplacement, deskilling, and wage compression.
• AI adoption proves labor demand rather than indicating the early installation of labor substitutes.

Social Function

Primary classification: transition management, ideological anesthetic, and partial truth.

The partial truth is that workers who learn to integrate immature AI into live workflows can gain temporary leverage. The anesthetic is the implication that continuous training restores the old employment bargain. It does not. It converts workers into permanently tested adapters, responsible for remaining employable while owners capture the productivity gains.

The article also serves the corporate training ecosystem. It reframes insecurity as opportunity, credential decay as lifelong learning, and structural substitution as a need for better coaching. Risk is relocated from the system to the individual.

The Verdict

On-the-job learning is returning as a deployment instrument and temporary lag defense—not as a revival of mass career security. Firms are training workers because the machine transition is incomplete. Once AI absorbs the contextual knowledge supplied through that training, the learning loop becomes a conveyor belt feeding the replacement system.

The article mistakes the boot sequence for the operating system. It describes humans being prepared to install their own substitutes.