When Apprenticeship Fades, Expertise Becomes the Scarce Resource

May 05, 2026

9 min read

6 views

78%

The strange coincidence of our age

What if two of the most important trends in modern life are telling the same story? One is happening in religion: younger people are far less likely to identify as Christian or pray regularly than older Americans. The other is happening in work: generative AI can make people faster, more flexible, and more productive, but it cannot magically turn a beginner into someone with deep judgment.

At first glance, these seem unrelated. One is about faith, the other about software. But together they point to a deeper cultural shift: we are getting very good at removing friction, yet increasingly bad at creating formation.

That distinction matters. Friction slows you down. Formation changes what kind of person you become. Modern systems, whether digital tools or social institutions, are optimized to reduce delay, lower barriers, and flatten learning curves. But the things that make a civilization durable, a profession trustworthy, or a faith tradition alive are not just speed and convenience. They are the slow accumulation of skill, habit, interpretation, and commitment.

The real question is not whether AI can help people do more. It can. The real question is: what happens when a culture gets excellent at assistance but weak at apprenticeship?

The hidden crisis is not decline, it is thinning

The religious data and the AI data both point toward a broader social pattern: expertise is becoming thinner at the same moment that access is becoming broader.

In religion, this can show up as a generation that knows less by participation than by inheritance. Older people often carry a dense web of practice: prayer routines, seasonal rituals, scriptural familiarity, moral vocabulary, and a sense of belonging that was reinforced over decades. Younger people may still admire some of the same values, but admiration is not embodiment. They have fewer repeated practices to turn belief into instinct.

In work, a similar thing is happening. A novice using AI can draft a report, write code, brainstorm a campaign, or summarize a market trend. The output may look polished. But polished output is not the same as grounded understanding. A person with real expertise does not only produce answers. They know which questions matter, which exceptions break the rule, and which elegant solution will fail in the real world.

This is why AI is so powerful and so limited at the same time. It can compress the distance between intention and first draft. It can make a beginner look competent earlier. But it cannot supply the invisible structure that expertise requires: pattern recognition, judgment under uncertainty, and the capacity to notice what the tool misses.

A society can automate competence signals faster than it can cultivate competence itself.

That is the hidden danger. When surface performance gets easier, we may confuse visibility with depth. We may start rewarding people for sounding competent, posting competent, or producing competent looking artifacts, while the underlying craft atrophies.

Why apprenticeship matters more when tools get better

There is a seductive story about technology: as tools improve, the need for human mastery declines. The real story is almost the opposite. Better tools often raise the premium on human judgment because they expand what is possible faster than they expand what is wise.

Think of a surgeon. A sophisticated machine can assist with precision, but it does not replace the surgeon's ability to read the room, sense a complication, or decide when the ideal procedure is not the right one for this patient. Or think of a jazz musician using digital tools. The tools can generate sounds, but they cannot improvise with taste. They cannot hear tension and resolve it emotionally. The tool amplifies the creator, but only if the creator already has something to amplify.

This is the key limitation of generative AI in knowledge work. It can lower the barrier to entry for many tasks, and that is genuinely valuable. A marketer can explore SEO optimization more quickly. A data scientist can step into a finance-adjacent role with less retraining. Teams can become flatter because more people can participate in adjacent domains.

But the tool does not erase the need for the deep structures that make those tasks meaningful. If you do not understand the domain, you cannot tell whether the model's confident answer is merely plausible or actually correct. You may be able to complete the assignment, but you will not necessarily become someone who can teach, adapt, or lead in that field.

This same distinction appears in religion. A culture can make belief easier to access through media, apps, clips, and personal branding. Yet easier access does not automatically create religious depth. Prayer is not just a behavior, but a discipline that trains attention. Belonging is not just a label, but a shared grammar of meaning. A faith tradition survives not by making entry frictionless, but by turning repeated practice into a form of vision.

The uncomfortable lesson is that formation requires repetition, constraint, and time, all three of which modern systems try to minimize.

The tool solves for output. Tradition solves for transformation.

This is the deepest connection between these two trends. AI is astonishing at generating output. Religion, at its best, is astonishing at generating transformation. One asks, “What can I produce right now?” The other asks, “What kind of person am I becoming?”

That difference may sound abstract, but it is the source of many current disappointments. We keep asking our institutions to provide outcomes without demanding formation. We ask schools to create workers, platforms to create communities, workplaces to create meaning, and apps to create discipline. Then we are surprised when everything becomes efficient and shallow at the same time.

Consider how people learn to cook. A recipe app can produce a meal. A video can show the steps. But the home cook who becomes truly good learns from failed sauces, burnt edges, timing mistakes, and the embodied memory of how ingredients change under heat. The app helps. The kitchen teaches.

Now compare that to the life of faith or the life of craft. A person may watch lectures about prayer or leadership, and even imitate the language. But without repeated practice, correction, and a community that notices when something is off, the learning stays fragile. They have information without initiation.

This is why both religion and work face the same modern temptation: to substitute frictionless access for thick apprenticeship. When that happens, participation may rise for a while, but resilience declines. People know how to begin, not how to persist. They can ask for help, but not internalize standards. They can imitate excellence, but not inhabit it.

The danger of easy access is not laziness. It is premature confidence.

Premature confidence is uniquely corrosive because it feels like progress. It produces visible artifacts, but not necessarily durable capability. A novice who can now draft like a professional may stop noticing the boundaries of their own understanding. A believer who can consume religious content may mistake exposure for devotion.

What a high AI, low apprenticeship culture will look like

If this pattern continues, the cultural effects will be subtle before they are dramatic.

First, organizations will produce more work products and fewer genuine experts. Teams will look productive because AI helps everyone contribute. But the number of people who can handle edge cases, mentor others, and make hard calls may shrink. In a crisis, that matters. The organization will appear wide but be structurally shallow.

Second, institutions will become more dependent on a small core of people who still have deep formation. In companies, that means the veterans who can distinguish signal from plausible noise. In religious communities, it means the few who actually know how to pray, interpret, counsel, and sustain a tradition through ambiguity. When those people are absent, the system becomes a shell.

Third, individuals will increasingly confuse access with assimilation. They will have more tools than ever to enter unfamiliar domains. That is a genuine democratization of opportunity. But the danger is that access can produce a false sense of mastery. Someone can move quickly into an adjacent role, or speak fluently about a tradition, without having undergone the slower processes that make fluency trustworthy.

This is where the analogy between AI and religion becomes more than metaphor. Both are tests of whether a society can preserve depth under conditions of speed. Can we use tools that compress learning without abandoning the practices that create wisdom? Can we make entry easier without hollowing out initiation?

That is not a technical question. It is a civilizational one.

The new literacy: knowing when to outsource and when to stay slow

The most useful response is not to reject AI or romanticize old forms of authority. It is to develop a new literacy: the ability to know which tasks can be accelerated and which must remain slow.

Use AI when the goal is exploration, first drafts, idea generation, or reducing repetitive load. Those are places where the machine is a force multiplier. But do not use it to skip the hard parts of expertise: struggle, diagnosis, revision, and judgment. Those are not inefficiencies. They are the curriculum.

The same principle applies to faith and other deeply human practices. Technology can provide reminders, access, and exposure, but it cannot replace the habits that train attention and character. A prayer app may help you remember to pray. It cannot make prayer into second nature if the deeper discipline never takes root. A streamed sermon may inspire you. It cannot substitute for the long work of belonging, service, confession, and shared ritual.

A good test is this: does the tool help you do a task, or does it help you become the kind of person who can do the task without the tool?

That question separates augmentation from dependency. It also separates formation from convenience. If a system merely makes performance easier, it may be efficient. If it makes capability durable, it is transformative.

Key Takeaways

Distinguish output from formation. A polished result is not proof of deep understanding. Ask whether a tool or practice is helping you create something, or helping you become someone.
Treat apprenticeship as a feature, not a delay. The slow parts of learning, repetition, correction, struggle, are not obstacles to success. They are the mechanism by which judgment is built.
Use AI for acceleration, not replacement. Let it shrink the first 60 percent of a task, but do not let it erase the hardest 40 percent, where standards, nuance, and expertise live.
Watch for premature confidence. When tasks become easier, people may overestimate their competence. Build in checks, review, and real-world feedback to keep confidence calibrated.
Protect high-friction practices. In work, that may mean mentoring, deliberate practice, and postmortems. In faith and life, it may mean ritual, silence, community, and repetition.

The real question: what are we optimizing for?

We like to say we want efficiency, growth, and access. But those are only means. The deeper question is what kind of human capacity we are trying to preserve.

A culture that only optimizes for friction reduction will eventually produce people who can begin quickly but struggle to deepen. A culture that only optimizes for tradition will risk stagnation and exclusion. The challenge of our moment is to combine accessibility with apprenticeship, speed with formation, and assistance with wisdom.

That is why the parallel between religious decline and AI productivity matters. Both reveal the same fault line in modern life: we are increasingly able to simulate competence without cultivating conviction, and to generate outputs without generating depth.

The future will not belong to those who can merely use tools or repeat inherited forms. It will belong to those who know how to be formed by practices that still matter, even when shortcuts are available.

In the end, that may be the most important skill of all: not just learning faster, but learning in a way that changes you.