Many professional careers begin with routine work: the research nobody wants to do, the first-pass document review, the market scan built by hand before anyone trusts you with the real analysis. It’s tedious, and it’s also how junior employees quietly build the judgment senior employees rely on without thinking about it. That entry point is disappearing. According to a McKinsey analysis published on July 14, 2026, AI is increasingly absorbing routine tasks such as research, documentation, data cleanup, and preliminary analysis. These activities have traditionally helped junior employees develop professional judgment, raising questions about how organizations will train their next generation of experienced workers.
Image Source: Pixabay
This dynamic extends beyond general concerns over entry-level hiring volume; it highlights a specific shift in how professional judgment is cultivated in the first place, and it connects directly to something Lorka AI’s research into the AI gender gap already flagged from a different angle: The issue connects directly to the gender gap in AI adoption explored in Lorka AI’s August 2026 interview with AI governance strategist Azahara Corrales. Citing McKinsey Global Institute research, Lorka reported that 33% of men in entry-level roles were encouraged by managers to use AI compared with 21% of women, making men nearly 1.6 times more likely to receive manager encouragement.
The article highlights how unequal encouragement can influence who experiments with AI, develops confidence, and contributes feedback to the teams designing these technologies.
Two Speeds, One Widening Gap
The data on who’s affected is specific. An August 12, 2026 update from Stanford’s Digital Economy Lab, based on ADP payroll records, found that employment among workers aged 22 to 25 in highly AI-exposed occupations was approximately 19% below where it would have been had it kept pace with employment among similarly aged workers in less-exposed occupations.
Experienced workers showed no comparable gap. The researchers found that the divergence appeared to be driven primarily by reduced hiring of younger employees rather than increased separations. However, they cautioned that the data do not establish how much of the change was caused by AI rather than other labor-market factors.
A separate July 2026 policy brief from Stanford’s Institute for Economic Policy Research found little evidence that AI was causing widespread job losses across the economy. However, the researchers acknowledged that AI may be contributing to a more difficult labor market for recent graduates, particularly in occupations involving routine research, analysis and writing.
Two senior Microsoft engineering leaders quoted in the same McKinsey report describe the split as an “AI boost” for senior staff, whose output AI multiplies because they already have the judgment to steer and check it, against an “AI drag” on junior staff, who often lack the experience to know when AI’s work is subtly wrong. The gap between what a senior employee can accomplish with AI and what a junior employee can safely contribute with it is widening, and the traditional bridge between the two, years of supervised repetition, is the thing quietly eroding.
Some Employers Are Rebuilding the Practice Work on Purpose
Image Source: Pixabay
A handful of companies are treating this as a design problem rather than an inevitability. McKinsey’s report describes a real estate firm that now has junior staff build market assessments the old way first, walking neighborhoods, studying traffic patterns, forming an independent judgment, before comparing that manual work against an AI-generated version to see exactly where the gaps are. The point isn’t nostalgia for slower work. It’s forcing juniors to develop independent judgment before they’re allowed to use AI as a shortcut, so the shortcut has something real to check itself against.
Bank of America has taken a similar approach at a larger scale. The bank announced that it would welcome nearly 4,000 summer interns and full-time campus recruits in 2026, maintaining its previous year’s intake while redesigning early-career roles around AI.
Its training approach includes AI-based simulations intended to help junior employees develop the judgment they would traditionally have acquired through routine work and supervised experience.
The bank’s head of global talent, Josh Bronstein, described the goal plainly: giving people real experience in a simulated way, quickly, rather than assuming AI exposure alone will teach it.
Practice Work Was Never Really About the Task
What both of those approaches share is a recognition that the old apprenticeship model wasn’t really teaching the task itself- walking a neighborhood, drafting a first-pass memo- was never the point. It was teaching pattern recognition through low-stakes repetition, the kind of judgment that only forms after you’ve done something wrong enough times to notice what wrong looks like. AI is extremely good at producing the finished output that repetition used to build toward. It’s not built to produce the repetition itself, and nobody designed a substitute for it before the shortcut arrived.
That’s a genuinely different problem than “AI is taking entry-level jobs,” and it demands a different response. Cutting junior headcount doesn’t solve it, because the problem isn’t a shortage of junior roles; it’s a shortage of structured ways for the juniors who are hired to build real judgment before they’re expected to supervise AI that’s already faster and more confident than they are.
Whoever Solves This First Gets the Next Generation of Senior Talent
The debate around this shouldn’t be framed only as fairness to new graduates, real as that concern is. It’s also a pipeline question for every organization watching senior staff outperform because of decades of accumulated judgment: If the tasks that once helped junior employees develop judgment continue disappearing, organizations will need more deliberate ways to build the next generation of experienced professionals. Structured training, supervised practice, and meaningful feedback could become increasingly important to maintaining that pipeline.
The challenge is no longer simply deciding which tasks AI can perform. It is ensuring that the employees expected to oversee those systems have opportunities to develop the judgment required to do so.
