AI Takes Over Entry-Level Work, Where Do Young People Get Started?

By: Tan Yinliang, China Europe International Business School

Sorting out materials, checking statements, and drafting plans are often the first lesson in young people's career. The work is not complicated, and sometimes even trivial, but it allows them to get in touch with the real business. Nowadays, AI can take over more and more tasks among them.

While enterprises save time and cost, young people may face a new problem: if those jobs that used to be used to accumulate experience no longer require human participation, where will their experience come from?

This is related to whether graduates can find their first job, and also related to whether enterprises can find the next batch of backbones in a few years. An industry cannot only need experienced people but no longer give opportunities to inexperienced people.

On October 23, 2026, the General Office of the State Council issued the Opinions on Adapting to the Development of Artificial Intelligence Technology and Helping Build an Employment-Friendly Development Mode, which arranged the employment risk assessment of AI application, joint training between schools and enterprises, and employment monitoring and early warning. To implement these arrangements, a problem worthy of attention is how to rearrange the career starting point of young people after technological progress.

The recruitment entrance may be narrowing quietly

When talking about the impact of AI on employment, people tend to think of layoffs. But for young people, changes may also happen in another way: enterprises do not lay off employees, they just no longer recruit new people as they did in the past.

Imagine a department with basically stable business scale. Originally, several assistants needed to be recruited to be responsible for sorting out information, making statements and preliminary analysis. Now, senior employees can complete a lot of work with the help of AI. Managers may decide to recruit fewer people, and existing employees will not lose their jobs. On the financial statements, this seems to be a reasonable efficiency gain; but for graduates who are looking for jobs, several opportunities to enter the industry have disappeared.

Such changes are easy to be ignored. The total number of employees can remain stable, and the department's performance may even grow. Only when it comes to campus recruitment, young people find that there are fewer positions suitable for them, and enterprises pay more attention to work experience.

Of course, we can not attribute the reduction of recruitment to AI. Market demand, economic environment and enterprise operation adjustment will all affect employment. What we need to continuously observe is which basic tasks are reduced after enterprises adopt AI, which new positions for new graduates change accordingly, and whether new growth opportunities have emerged.

The Ministry of Education previously predicted that the scale of national ordinary college graduates in 2026 will reach 12.7 million. For such a huge group, it is not enough to just know how many positions are in the market, and we also need to know how many of them are willing to accept and train new people.

In the current autumn recruitment, Meituan's official recruitment website has listed positions for fresh graduates such as AI full-stack engineer and AI back-end development engineer. New opportunities are emerging. We also need to care about how more graduates can start from no experience and reach the day when they can take independent responsibility.

Today's novices are tomorrow's backbones

Junior work has an easily overlooked value: it is not only completing tasks, but also training people.

A new finance employee who checks vouchers learns not only to match the numbers, but also why the same business has different calibers and which anomalies need to be followed up. A sales assistant who follows up orders gradually understands why customers rush for delivery and why procurement and production are difficult to coordinate. These judgments can hardly be mastered by just listening to a lecture.

Work experience is often accumulated in the process of doing things repeatedly, making mistakes, being reminded, and then revising. Basic tasks can be automated, but this learning process needs to be redesigned.

If the company hands over all simple work to AI and only leaves complex problems to new employees, new difficulties will arise. If a person is not even familiar with the normal business process, how can he judge whether the AI's suggestions are reasonable? Without dealing with the needs of ordinary customers, how can he directly deal with complex complaints? Arranging new employees to review AI outputs also requires corresponding training for them.

For enterprises, training fewer new employees may soon be reflected in higher per capita efficiency. The loss of talent reserve will not appear on the current statements immediately. When backbones leave, the business expands, or someone needs to take over, it is found that there are not enough people who can take independent responsibility.

If every enterprise hopes to recruit mature talents from elsewhere but is unwilling to bear the training cost, the talent supply of the whole industry may be affected. An expense that seems to be saved today may reappear in a few years in the form of higher recruitment costs and longer running-in period.

This change will also widen the opportunity gap among young people. People with family resources and internship channels can accumulate experience in advance; those who lack these conditions rely more on junior positions in open recruitment. The rising experience threshold may also turn the gap that could have been made up through work into an obstacle in job hunting in advance.

Let AI help young people develop real skills

AI can also lower the threshold for new employees to learn. The basic knowledge that used to require repeated consultation with colleagues can now be understood with the help of AI first; the preliminary analysis that was difficult to complete independently in the past can also be made into a version first, and then the mentor can point out the problems.

Opportunities are not only for people who develop large models. Xinhua News Agency reported in June this year that Guo Haifeng, a student of Xi'an University of Science and Technology, studied the relevant "micro-major", combined mining engineering with artificial intelligence and big data technology, and found a satisfactory job. This example reminds young people that learning AI can start from their own major and find specific problems that tools can help solve.

Taking inventory management as an example, after AI gives a restocking suggestion, students or new employees still need to check whether the sales data is affected by promotions, whether suppliers can deliver goods on time, and whether the warehouse can accommodate new inventory. Only by clarifying these conditions can we judge whether the suggestion is feasible. Learning takes place in the process of verification and correction.

Enterprises can let new employees put forward preliminary judgments independently first, then compare them with AI results, and senior employees explain the differences; they can also use historical cases for training, and gradually expand the scope of responsibility after new employees can explain why mistakes occur. AI speeds up the speed of getting answers, while mentors need to help new employees understand the conditions under which the answers are valid.

School assessments should also change accordingly. A well-formatted, fluent report is increasingly difficult to show what students have actually learned. Teachers can ask students to explain the basis in person, temporarily change a business condition, and then see if they can adjust the plan. Being able to explain, correct errors and respond to changes is closer to the requirements of real work.

For young people, writing several more AI tool names on their resumes has limited effect. What is more convincing is a complete project: what problem you faced, what information you verified, what mistakes you found and corrected, and what results you finally produced. Enterprises ultimately need someone to be responsible for the work.

Someone needs to take new employees seriously for their first job

To put these ideas into practice, enterprises first need to re-examine positions for new employees. Positions can change with technology, but opportunities for real business, mentor guidance and gradual assumption of responsibility should be retained.

For example, during the 6 to 12-month training period, new employees can gradually move from being familiar with processes and participating in projects to independently being responsible for tasks in a limited scope. The mentoring time should be included in the workload of backbone employees. If supervisors only assess the current output, no one can spare time to train new employees seriously.

Enterprises can also use part of the time saved by AI for businesses that they could not take care of in the past, such as following up more small and medium-sized customers and improving after-sales service, and then let new employees participate in them. Whether efficiency improvement can create positions is also restricted by market demand, but it is worthwhile for enterprises to calculate more accounts of business expansion and talent training.

The cooperation between schools and enterprises needs to provide continuous practical opportunities. The Opinions proposes to promote students to practice in enterprises and enterprise engineers to teach in schools. The two sides can jointly develop real projects that last for a semester, so that students can revise their plans repeatedly and see why their suggestions are accepted or rejected. To evaluate internships, we should see what tasks students have undertaken and what feedback they have received.

Public policies can also make support more specific. The Opinions has required to carry out special investigations on the employment impact of AI applications. In the implementation process, we can continuously track the number of positions for fresh graduates, the proportion of positions requiring zero experience, the duration of first job hunting and the situation of probation to regular employment. When the recruitment of new employees in certain industries continues to decrease, the reasons should be further investigated and support should be provided in time.

The 15th Five-Year Plan for Implementing the Employment Priority Strategy issued by the State Council has proposed to implement the development plan of one million employment trainee positions. On this basis, we can select some positions to pilot human-machine collaborative training, link part of the subsidies with mentor input, ability improvement and subsequent employment situation, and combine with position difficulty evaluation to encourage enterprises to train new employees seriously.

For small and medium-sized enterprises lacking training resources, industry associations, public employment agencies and universities can jointly provide project and mentor services. High-quality practical opportunities should also be prioritized for graduates from poor families and those who lack internship channels. Whether one has the resources to find the first internship should not prematurely determine a person's career starting point.

Young people need to learn AI actively, and enterprises and schools also need to provide scenarios where they can learn. It is difficult to solve the problem by individual efforts if graduates are only required to become experienced as soon as possible but no one gives them the opportunity to accumulate experience.

AI will constantly change the content of work. Whether an enterprise has enough people who can take charge independently in the future also depends on whether it is still willing to train new employees seriously today. Arranging the first job well for young people is to create growth opportunities for the next batch of backbones.

The author is a professor of Decision Sciences and Management Information Systems at China Europe International Business School