After AI takes over all entry-level work, how are we supposed to cultivate new talents?

In the past, when an enterprise trained a new employee, there was usually a relatively clear growth path: starting with simple work, gradually getting familiar with the business, then taking on complex tasks, and finally developing the ability to solve problems independently.

A newly joined engineer in a manufacturing enterprise may start by sorting out drawings, compiling technical documents, collecting quality data and analyzing simple problems; a junior consultant may initially be responsible for data collection, meeting minutes, process sorting and drafting preliminary reports. Although these tasks are basic, they give new employees the opportunity to get in touch with real business, understand professional terminology, and gradually build their own judgment ability.

But now, AI is completing these tasks rapidly. It can sort out materials, generate reports, analyze data, write code, and even put forward preliminary solutions. Enterprises have found that tasks that used to take junior employees several days to complete can now be finished by experienced employees with the help of AI in just a few hours.

This seems to be a good thing. Enterprises have improved efficiency, and employees have also reduced repetitive labor.

But a new problem has emerged:

If all junior tasks are handed over to AI, through what channels can new employees accumulate experience? If enterprises no longer need so many junior employees, where will the senior talents come from in five or ten years?

This may be a deep-seated challenge for enterprise talent cultivation in the AI era.

Junior tasks not only generate output, but also undertake the function of talent cultivation

In the past, enterprises arranged for new employees to start with basic work, ostensibly to complete some relatively simple tasks, but these tasks actually undertook another important function: allowing new employees to learn in real work.

For example, a quality engineer who has just joined an enterprise may first be responsible for sorting out records of non-conforming products, counting defect types, and drawing trend charts. As the work progresses, he gradually finds that the same product defect may come from different process reasons, and the same process abnormality may show different quality problems.

These understandings do not entirely come from training materials, but are formed in the process of continuously contacting actual data, on-site problems and empirical judgment.

The same goes for R&D engineers. At the beginning, they may only modify drawings, sort out BOMs, and check dimensional tolerances, but when dealing with these tasks repeatedly, they begin to understand the relationship between materials, structures, manufacturing processes and product functions.

The value of junior tasks is not only to complete the tasks, but also to allow new employees to gradually build their perceptual ability to professional problems.

What AI has changed is precisely this part. It can quickly complete a large number of basic tasks with clear rules, information intensity and repeatability, but it may also deprive new employees of the opportunity to access business details through these tasks as they did in the past.

This means that the original talent cultivation mechanism of enterprises may need to be redesigned.

AI may allow new employees to skip work steps, but they cannot necessarily skip ability growth

Imagine such a scenario.

A newly recruited quality engineer receives a task to analyze the recent fluctuation of non-conformity rate on a certain production line. In the past, he needed to collect inspection data, sort out defect categories, draw control charts, understand the on-site process, and consult experienced engineers about the cause of the abnormality.

This process may take two days, but he can gradually understand the relationship between data and the manufacturing process.

Now, he hands the data to AI, and gets an analysis report in a few minutes, including trend changes, possible causes and improvement suggestions.

From the perspective of work efficiency, this is an obvious progress.

But if you directly ask him: "Why do you judge this is a special cause fluctuation?" "Which data supports this conclusion?" "If AI's judgment is wrong, how will you find it?"

He may not be able to answer.

The problem is not that AI makes new employees learn too fast, but that new employees may get seemingly very professional answers before they understand the problems.

In the traditional working mode, a person's work output is often strongly related to his professional ability. If a new employee cannot produce a complex analysis report, it usually means that he lacks the corresponding knowledge and experience.

But in the AI era, this relationship is being broken. A person with insufficient experience can also use AI to generate a report with complete structure, professional language and exquisite charts.

As a result, it is more difficult for managers to judge the real ability of employees only by work results.

This also brings new risks: enterprises seem to have more and more employees who can complete complex tasks, but how many of them truly have the ability of independent judgment and problem solving?

Enterprises may be losing the opportunity to cultivate future experts

The impact of AI on junior positions has begun to attract the attention of research institutions.

In a study updated by the Stanford University Digital Economy Laboratory in August 2026, it was found that in occupations with high AI exposure in the United States, the employment performance of young employees aged 22 to 25 is significantly weaker than that of their peers in occupations with low AI exposure. The study also pointed out that there has not been a universal job substitution in the overall labor market, and not all of these employment changes can be attributed to AI.

This study reminds us that the changes brought by AI may not be that all positions are reduced at the same time, but that junior positions and experienced positions are affected differently.

In the past, the talent echelon of enterprises usually started from the cultivation of junior employees, and after years of practice, they gradually grew into senior engineers, technical experts and managers.

But if enterprises are increasingly inclined to let senior employees directly complete work with the help of AI, and reduce the recruitment and cultivation of new employees, they may reduce labor costs in the short term, but weaken the internal talent supply in the long run.

For example, an enterprise believes that two senior engineers with the help of AI can complete the work that used to require five people, so it decides to reduce the recruitment of three junior engineers.

In the short term, this adjustment may increase per capita output. But five years later, when senior engineers retire, leave their jobs or the enterprise needs to expand its business, who will undertake more complex engineering decisions?

Enterprises save the cost of cultivating new employees today, but may increase the risk of shortage of professional talents in the future.

This does not mean that enterprises should retain all low-efficiency jobs just to cultivate new employees, but they need to realize that talent cultivation itself is an organizational activity that requires continuous investment.

In the past, this investment was hidden in daily work allocation. Now, when AI takes over more and more basic tasks, enterprises must reconsider how to consciously cultivate new employees.

The growth path of new employees needs to shift from "completing tasks" to "understanding decision-making"

Since AI can complete a large number of basic tasks, should enterprises still require new employees to repeat the past learning process?

The answer is not necessarily.

For example, today when computers are widely used, there is no need for us to require engineers to manually complete a large number of complex calculations every day. But engineers still need to understand the calculation principles, input conditions and applicable boundaries of the results.

Talent cultivation in the AI era should also follow a similar logic.

What needs to be retained is not the inefficient working methods, but the indispensable professional understanding and judgment training behind these tasks.

In the past, quality engineers may gradually understand root cause analysis, corrective measures and effectiveness verification by compiling dozens of 8D reports.

In the future, AI can assist in report writing, but new employees still need to participate in on-site investigation, evidence collection, cause judgment and measure verification.

In the past, R&D engineers accumulated design experience by modifying drawings repeatedly. In the future, AI can automatically complete part of drawing inspection and model modification, but new employees must understand why the design scheme is valid, why some structures may fail, and how to choose between different technical schemes.

Therefore, the cultivation method needs to change: from training employees how to complete each operation step, to gradually training them how to understand problems, put forward hypotheses, judge risks, select schemes and verify results.

This does not mean that basic knowledge is no longer important. On the contrary, only those who master the basic principles have the ability to judge whether the AI output is reasonable.

In the AI era, the role of senior employees also needs to change

In many enterprises, the traditional mentoring cultivation mainly relies on new employees to follow senior employees to participate in actual work. The mentor is responsible for complex problems, and the new employee takes on auxiliary tasks, and gradually grows through observation, questioning and repeated practice.

But when a large number of auxiliary tasks are handed over to AI, this naturally formed learning opportunity may decrease.

Therefore, enterprises need to redesign the cooperation mode between senior employees and new employees.

For example, in a product design review, in the past, young engineers may prepare materials, and senior engineers are responsible for judgment. In the future, AI can complete most of the material sorting, and new employees should have the opportunity to analyze design risks in advance, put forward their own judgments, and then senior engineers explain why a certain design scheme is selected.

In a quality problem analysis, new employees can first independently put forward possible causes, then use AI for supplementary analysis, and finally senior engineers discuss it combined with on-site evidence.

The key is not to let new employees follow AI to learn standard answers, but to let them understand how experienced engineers deal with uncertainty, how to identify key evidence, and how to make choices among multiple schemes.

An important responsibility of senior employees in the future may not only be to solve the most complex problems, but also to transform their professional judgment process into experience that new employees can learn from.

This also puts forward higher requirements for the enterprise's knowledge management. In the past, many expert experiences existed in individual minds, and enterprises only saved the final work results. In the future, enterprises need to pay more attention to the precipitation of key technical decisions, problem analysis processes, and experiences and lessons.

The talent evaluation method cannot only focus on work results

After AI becomes popular, enterprises also need to rethink how to evaluate new employees.

In the past, a new employee who could independently compile a complete analysis report might be regarded as an important manifestation of ability improvement. But now, AI can help him quickly complete a seemingly very mature report.

If enterprises still evaluate employees only according to report quality, completion speed and number of tasks, they may overestimate the actual ability of some people.

In the future, in addition to work output, attention should also be paid to whether employees understand the business logic behind the tasks, whether they can find errors in AI analysis, whether they can explain the basis of their own judgment, and whether they can put forward reasonable solutions when there is no standard answer.

For example, for a quality engineer, we should not only evaluate whether he can complete PFMEA, but also evaluate whether he truly understands the failure mechanism, can identify key risks, and can judge whether the existing prevention and detection measures are effective.

For an R&D engineer, we should not only evaluate whether the drawings are correct, but also pay attention to whether he can explain design choices, identify manufacturing risks, and put forward reasonable schemes when requirements change.

AI can help employees improve the level of work output, but enterprises need to establish mechanisms to identify whether employees' own abilities are improved synchronously.

Otherwise, enterprises may cultivate a group of employees who are good at using AI to complete tasks but lack independent professional judgment ability.

What really needs to be redesigned is the enterprise's talent cultivation mechanism

Talent cultivation in the AI era should not simply return to the past, nor can it completely rely on new employees to learn by themselves.

Enterprises can consider establishing a new cultivation mechanism: retain necessary basic training, let new employees participate in real business as early as possible, improve learning efficiency through AI, and at the same time build professional judgment ability through the guidance of senior employees.

For example, in the initial stage of new employee cultivation, it is not necessary to require them to independently complete all document and data sorting work, but appropriate basic practice should be arranged to ensure that they understand the core principles.

In the stage of ability improvement, new employees can use AI to analyze real cases, but they are required to explain the analysis logic, identify key assumptions, and verify the results through on-site investigation or actual data.

In a higher stage, new employees can be allowed to take charge of complete small projects, and be responsible for problem analysis, scheme selection, risk control and final results.

The focus of this cultivation method is no longer to accumulate the number of tasks according to working years, but to gradually build the ability to solve problems independently through designed practice.

At the same time, enterprises need to set aside clear guidance time for senior engineers, instead of thinking that with AI, new employees no longer need mentors.

AI can shorten the time for knowledge acquisition, but it cannot automatically replace the experience formed by real practice, professional feedback and responsibility taking.

We should not let AI improve today's efficiency but weaken tomorrow's capabilities

AI is changing the division of labor in enterprises, and is also changing the growth path of professional talents.

In the past, junior tasks were not only part of enterprise production activities, but also an important way to cultivate the next generation of professional talents.

Now, when more and more basic tasks are taken over by AI, enterprises can no longer assume that new employees will grow naturally in their daily work.

This means that talent cultivation needs to change from relying on the natural accumulation of work experience in the past to more conscious capacity building.

What enterprises really need to solve is not how to let new employees continue to do the work that AI can already complete, but how to let new employees, with the help of AI, quickly form the professional judgment and responsibility ability that AI cannot take over for them.

If enterprises only focus on how many employees AI can reduce and how much cost it can save, they may obtain short-term efficiency, but ignore the construction of long-term talent echelon.

Only those enterprises that can redesign work, knowledge and talent cultivation mechanisms can obtain both the efficiency advantages brought by AI and the organizational capabilities for sustainable development.

The real competitiveness of enterprises in the future is not just how many people they have who can use AI, but whether they can continuously cultivate people who truly understand the business, can make judgments and make decisions.

This article is from the WeChat official account "TPP Management Consulting", author: TPP Management Consulting, authorized for release by 36Kr.