When AI takes over the tasks newcomers used to learn on, what keeps juniors building real skills — and is anyone protecting that?

This explores whether anything (laws, professional bodies, workplace practices) keeps people learning skills on the job when AI takes over the tasks newcomers used to learn on. The direct answer from this corpus is that it describes the problem in detail but documents no institution built to solve it. None of the retrieved work studies a guild, training mandate or company policy that protects apprenticeship. That gap is itself a finding, and the material around it shows why the gap matters.

The clearest picture of the threat comes from interviews with South Korean software engineers. Generative AI did not simply speed up junior work. It moved the entry-level tasks into senior engineers' AI workflows, so juniors lost the hands-on struggle through which expertise used to form Does generative AI prevent juniors from getting entry-level work?. Seniors and juniors also saw the problem differently, so the people best placed to protect the learning path may not notice that it is gone. Two findings suggest this is a pattern and not one country's quirk. Delegated AI use concentrates in information-heavy jobs, where much training happens through doing the work Where have workers actually delegated tasks to AI?. And the firms most exposed to AI replace outside workers fastest Do firms substitute labor for AI at different rates?. Learning paths may therefore shrink unevenly, firm by firm, with little outside visibility.

You might hope that using AI to augment workers, not replace them, solves this. A mapping of over 8,000 workplace AI risk scenarios found it doesn't. Heavy reliance on AI agents can slowly wear down worker skills, and with them the ability to oversee the AI at all Does AI augmentation protect workers from skill erosion?. This is where apprenticeship stops being only a career issue. One argument holds that society stays aligned with human interests partly because institutions depend on skilled people who care about outcomes. As that dependence fades, the informal check fades with it Does incremental AI replacement erode human influence over society?. Seen this way, losing apprentices also means losing future overseers.

Why is this hard for institutions to fix? Expertise is not just what a person knows. Communities grant it through participation and a track record Can AI ever gain expert community trust through participation?. Knowledge has also historically traveled with people: the teacher, the master, the giver. AI-mediated knowledge flows without those human carriers Is AI returning knowledge to flow-based economies?. Apprenticeship is exactly the kind of embodied transmission that a flow without carriers skips over.

The corpus offers indirect clues about what a protective institution would need. Debates over AI governance keep reaching the same conclusion: voluntary company commitments fail without government enforcement behind them Can companies alone manage the risks of AI systems? Can industry self-regulation slow AI without government enforcement?. A study of a long-running AI agent found that rules written into the environment the agent actually consulted worked better than policies kept on the side Can governance rules embedded in runtime memory actually protect autonomous agents?. Carried over to apprenticeship, these point the same way. Protection probably has to be built into how work is assigned, for example by reserving some tasks for juniors to struggle through, and backed by something with authority. A mentoring guideline on file is unlikely to be enough. These are inferences from neighboring debates, though, not findings about training policy. Studies of what actually protects skill development are a gap in this collection.