Employers Want Experienced Workers, but They Are Automating the Work That Once Produced Experienced Workers.

You can eliminate a position. You still have to account for the capability that position was supposed to produce.

You want someone who understands the business, recognizes a bad assumption, knows when the numbers are bullshit, and can make a decision without somebody holding their hand.

Fantastic. Where the fuck do you think that person comes from?

Somewhere between knowing nothing and knowing enough to be useful, somebody has to be allowed inside the operation. They have to see a decision get made, understand what it costs, watch the result, and learn why the explanation in the meeting didn't survive contact with the business. The U.S. Department of Labor's registered-apprenticeship model formalizes the same underlying mechanism through paid work, structured on-the-job learning, mentorship, and related instruction. [1]

Now suppose you automate the assignments that gave them that access, remove the junior positions, and keep demanding experienced applicants.

You haven't eliminated your dependence on experience. You've decided somebody else should pay to produce it.

That might be a defensible business decision. But own the fucking decision.

I want to know what you actually eliminated

My approach starts with diagnosis. Before I recommend spending money, I want to understand the operation that is supposed to benefit from it. I apply the same standard to saving money.

“We automated the department” tells me almost nothing.

Which tasks disappeared? Which decisions still require a person? Who checks the output? What happens when the information is incomplete? Who understands the relationship between the report, the customer, the budget, and the work somebody else has to deliver?

Show me the workflow before you show me the savings slide.

A beginner cleaning up customer records might be doing tedious work. They might also be discovering that marketing, intake, sales, and finance use the word “customer” to describe four different events.

Automating the cleanup could be useful. Automating it without anybody understanding those definitions could simply make the wrong information travel faster.

I don't give a shit whether the spreadsheet was completed manually. I care whether somebody understands what the numbers mean and what management is about to do with them.

That is the development question buried inside the automation decision.

Clicking buttons isn't the experience I'm buying

I specialize in high-ticket service businesses in competitive local markets. In that kind of business, knowing how to launch an advertisement is only one piece of understanding how the company makes money.

What happens after the inquiry? Can the customer afford the service? Do they actually want it? Can the business deliver it? What does delivery cost? When does the company collect?

An employee who sees only the advertising account can become very good at the advertising account while remaining dangerously uninformed about the business.

An employee who follows the opportunity through intake, sale, delivery, and collection has a chance to learn something more valuable: where a department's apparent success becomes somebody else's expensive problem.

That's the exposure I want a development system to provide.

Keeping somebody busy for three years doesn't prove you developed them. Neither does giving them AI and telling them to figure it out.

You have to let them see the fucking consequences.

Let's price the whole decision

Here's a hypothetical, not a client story or a salary benchmark.

A department has one senior employee earning $140,000 and two junior employees earning $60,000 each. Management believes AI will let the senior person produce the current output without the junior positions.

The apparent annual salary reduction is $120,000, before considering benefits, payroll costs, tools, transition expenses, or the cost of replacing any capability the junior positions provided.

Fine. Now finish the analysis.

How much additional review does the senior person perform? What software, implementation, security review, and maintenance does the new arrangement require? Does the senior person have time to improve the business, or are they now supervising a machine while handling every exception themselves?

What happens when that person leaves?

Can someone else operate the system? Is the process documented? Can the company afford an experienced replacement? How long could it carry reduced output while that replacement learns the business?

The juniors weren't guaranteed to become successors. They might have left. They might have needed more supervision than the company could justify. Those are real considerations.

But so is the risk of building an operation around one person and assuming that person will remain available at the price you prefer.

I would compare the alternatives: automate and retain a development role, redesign the junior work, cross-train another employee, purchase outside capability, or accept the dependence and reserve capital for it.

The positions may disappear. The capability requirement does not disappear merely because it moved somewhere less visible.

Management chooses the risk. My objection is to presenting a partial calculation as the entire fucking answer.

You don't get training for free either

Retaining junior employees is not enough.

If the senior employee is measured entirely on personal output, then every hour spent teaching competes with the number management uses to judge them.

You can announce that mentorship matters. The calendar will tell everybody whether you mean it.

Who reviews the beginner's work? What authority can the beginner exercise? Which mistakes are survivable? When does responsibility increase? What work gets removed from the senior person's schedule so they can teach properly?

Someone has to own that system. Someone has to fund it.

And the beginner needs to explain their reasoning. I want to hear what information they used, what they couldn't establish, what assumption matters most, and what would change their recommendation.

A beautiful answer they can't defend tells me they can produce a beautiful answer. That's all it tells me.

AI can accelerate the learning if you design the learning

Let a junior employee use AI to organize records, draft an analysis, or compare possible explanations. Then make them trace the conclusion back to the business.

Suppose the draft recommends increasing advertising because inquiries are down. Before anybody spends a dollar, I want the employee to investigate whether the issue is demand, tracking, qualification, response time, capacity, or an offer the market doesn't want.

That exercise teaches diagnosis. Copying the recommendation into a presentation teaches dependence.

Access has to match responsibility, too. A learning assignment doesn't authorize dumping confidential customer records or internal financial information into whatever tool produces the fastest answer. Use information and systems approved for the work. The employee can learn the operating logic without receiving every sensitive detail in the company.

AI can help with the assignment. A person still has to establish whether the answer fits reality and who can act on it. NIST's voluntary AI Risk Management Framework similarly treats defined human roles, oversight, context, privacy, and evaluation as matters organizations have to govern rather than assume away. [3]

Your hiring complaint belongs in your capital discussion

If you choose to buy experience instead of developing it, budget for buying it.

That includes compensation, recruitment, onboarding, and the possibility that a capable applicant has choices. You don't get to eliminate development costs and then demand that the labor market sell you the finished product at an entry-level price.

If you choose to develop people, fund supervision and give them meaningful exposure. If the business can't afford either approach, adjust the work and the objectives around that constraint.

That's how I think about capital elsewhere in the operation. Approve a smaller investment and the expectations have to change. Calling the constraint “AI efficiency” doesn't exempt it from scrutiny.

I am not claiming every eliminated junior position creates a talent shortage. Some tasks teach very little. Some businesses can reliably hire the capability they need. Some tools may reduce the amount of human expertise a particular process requires. The Bureau of Labor Statistics likewise cautions that occupational exposure to AI does not itself establish whether employment in that occupation will grow, decline, or be automated. [2]

Those facts could justify the decision. Investigate them.

But if your business still needs human judgment, show me how you intend to obtain it next year and five years from now.

Don't show me the payroll reduction and expect applause before we've discussed what stopped being built.

You want experienced workers? Then account for the cost of experience.

It doesn't become free because you stopped producing it.

Continue Reading on anuhlest

• You Hired Someone. Now What the Fuck? — why hiring has to begin with the capability, authority, resources, and continuing need the business actually requires.
• Should I Hire a Marketing Agency, an Employee, a Consultant, or Build It In-House? — how the nature of the work should determine the working relationship.
• The Importance of Animal — why AI can help examine available knowledge without replacing direct observation and experimentation.