The most awkward generation in the age of AI may be the newly graduated college students.

Recently, a Fortune 500 US-invested enterprise we connected with recruited two entry-level positions. The roles were not senior or management-level, but shortly after the job posting was released, it received approximately 1,000 resumes from fresh graduates. Even more surprisingly, around 500 of these applicants graduated from 985/211 universities or top 50 global higher education institutions. In the end, the enterprise only hired two candidates as planned.

The fact that 1,000 people are competing for 2 positions cannot simply be attributed to AI. The economic environment, job market, number of graduates, and the inherent attractiveness of large enterprises all contribute to this figure. But this case still makes me think of an ongoing change: in the AI era, the most awkward group of people in professional competition may be precisely the young people who have just stepped out of university campuses.

Because AI has different impacts on people at different career stages. For someone who has worked for 20 years, AI may be a capability amplifier; but for a new graduate who has not yet built up professional experience, AI may first become their competitor.

Why were enterprises willing to recruit so many young people in the past?

Looking back a dozen years ago, after a university graduate joined an enterprise, they usually did not take on particularly complex tasks immediately. They might start by organizing Excel sheets, generating reports, looking up materials, drafting meeting minutes, collecting data, making PPTs, sorting out customer information, tracking project progress, or helping senior engineers complete some basic analyses.

These tasks may seem low in technical content and even somewhat tedious. But they used to play a very important role: they served as the entry point for young people to step into the real business world.

A newly graduated quality engineer might:

- First sort out complaint data, then gradually get in touch with 8D methodology;

- First organize SPC data, then gradually understand why processes fluctuate;

- First accompany senior engineers to supplier sites, then slowly learn what to focus on during audits.

A young consultant might also start by sorting out interview transcripts, looking up industry materials, and making PPTs, and only after several years can they gradually analyze client problems independently.

Therefore, past career growth was essentially a long path: do simple tasks first → get exposed to real problems → keep making mistakes → get corrected by seniors → accumulate cases → form experience → finally develop judgment.

Enterprises recruit young people partly because their labor cost is relatively low, and partly because a large number of basic tasks do need to be completed by someone. While young people help enterprises finish these tasks, they also complete their own professional training. But AI is changing this exchange relationship.

The first tasks AI replaces are precisely young people's "entry-level jobs"

Look at what AI is best at doing now: looking up materials, organizing information, drafting meeting minutes, data processing, first drafts of reports, PPT frameworks, emails, translation, basic coding, industry research, scheme comparison, knowledge retrieval... A closer look will reveal that these are exactly the main job contents of many entry-level positions in the past.

1. In the past, a department head might need several assistants to help collect materials, sort out data and write reports. Now with the help of AI, they can complete a large part of the work on their own.

2. In the past, a senior engineer might need young engineers to help check standards, organize data and generate reports. Now many basic tasks can be finished in a few minutes.

3. In the past, consulting projects required young consultants to spend a lot of time sorting out interview transcripts, searching for industry materials, and building PPT frameworks. Now AI can complete a large part of this work.

This raises a very practical question: when an enterprise finds that one senior employee equipped with AI can complete the workload of a small team in the past, why do they still need so many entry-level positions?

AI may not directly "snatch" the job of a specific university graduate, but it is reducing enterprises' demand for a large number of entry-level execution tasks. And this is happening at a time when the number of university graduates is increasing, educational backgrounds are getting higher, and employment competition is becoming fiercer. As a result, the scenario we saw earlier may emerge: two non-senior positions attract 1,000 fresh graduates, a large number of whom are "top university students" in the traditional sense.

AI may have completely opposite effects on experienced people

Interestingly, the same AI can produce completely different effects on someone who has worked for 20 years. In the past, a quality manager with rich experience might be very familiar with site operations, suppliers and clients, but they might not be good at data analysis, not know how to write programs, read English materials slower than young people, and spend more time making PPTs and sorting out massive information.

All these used to limit their capability boundaries, but after the emergence of AI, these limitations are rapidly diminishing. They can ask AI to help sort out data, search for materials, compare standards, translate foreign languages, generate report frameworks, and even help write simple programs.

Tasks that used to require several young employees to assist in completing can now be finished by one senior professional with the help of AI. So a very interesting change has taken place: AI has not devalued their 20 years of experience, but instead amplified this 20 years of experience.

Because the real difficult problem is never "finding ten pieces of information", but knowing which three pieces of information are truly important. It is never generating five solutions, but knowing which one can actually be implemented on the factory site. It is never letting AI analyze a pile of data, but knowing whether the problem behind the data lies in equipment, materials, craftsmanship or management.

These capabilities can be assisted by AI, but it is very difficult for AI to build them for a person out of thin air.

What is truly appreciating in the AI era may be "judgment"

This is why I believe that the real dividing line in future professional competition may increasingly not be "whether you know how to do it", but "whether you know how to make judgments".

1. AI can help a quality engineer generate an 8D report, but it does not know whether the real root cause on site is the one written in the report;

2. AI can help procurement personnel compare suppliers, but it does not know whether the entire management team of a certain supplier, despite the boss's good promises, actually has mass production capabilities;

3. AI can help R&D personnel generate several technical solutions, but the final choice of solution requires comprehensive consideration of cost, reliability, manufacturing capacity, customer demand and project risks.

At this point, human value starts to shift from "producing answers" to "judging answers", and judgment largely comes from experience. How many failed projects you have seen, how many customer complaints you have handled, how many supplier sites you have visited, how many equipment abnormalities you have experienced, how many mistakes you have made, and after how many mistakes you truly understand the underlying reasons.

So to be more precise, AI is reducing the scarcity of execution capabilities, but increasing the value of judgment capabilities. This is why some professionals aged 40 or even over 50 may become beneficiaries of this round of AI development. They have accumulated a large amount of professional experience in the past, and AI suddenly helps them make up for their shortcomings in information acquisition, expression, data processing, language, and even some technical tools. For the first time, experience can be amplified at such a low cost.

Age itself does not automatically become an advantage

Of course, this does not mean that "the older you are, the more advantages you have in the AI era". A person who has worked for 20 years does not necessarily have 20 years of effective experience; they may just have repeated the same year of work 20 times. If they only complete tasks according to procedures for a long time, do not form an understanding of the business, do not accumulate methods to solve complex problems, and do not develop independent judgment, then AI may also impact their positions.

So what really matters is not being 40 or 50 years old, but what a person has precipitated after long-term work:

- If what you mainly have is execution capability, AI may become your competitor;

- If what you have is experience, professional knowledge and judgment capability, AI is more likely to become your amplifier.

This may be the real professional dividing line in the AI era.

Then comes the hardest problem for young people — without experience, how to develop judgment?

Here comes a paradox: we say that the most valuable asset in the future is judgment, and judgment comes from experience. But a fresh university graduate is precisely the group that lacks experience the most.

Worse still, the tasks that used to help young people gain experience are now largely being completed by AI. In the past, a young engineer might personally sort out 100 quality reports, and gradually learn which problems occur frequently in this process. They might follow senior technicians to dozens of site visits before knowing where to check first when equipment malfunctions; a young consultant might sort out dozens of interview transcripts before gradually understanding the difference between the "problem" mentioned by enterprise managers and the real problem.

Now AI can finish all these tasks in a few minutes, which undoubtedly improves efficiency, but young people's learning process may also be skipped together. This may be the most noteworthy problem in the AI era: AI may not only reduce entry-level positions, but also reduce opportunities for young people to build professional experience.

Enterprises will also face a bigger problem — where will senior talents come from in the future?

If more and more enterprises adopt the same strategy: "Let AI do entry-level tasks, and we only recruit experienced people", it is of course very cost-effective in the short term. But what about five years later? Ten years later?

The reason why an R&D engineer with 10 years of experience is available for recruitment by enterprises today is that ten years ago, there was an enterprise willing to hire an inexperienced university graduate and allow them to learn from simple tasks.

If the whole society starts to reduce entry-level positions while still expecting the talent market to continuously provide "experienced people", this is inherently unsustainable. Therefore, AI may eventually force enterprises to rethink their talent cultivation methods.

In the future, when enterprises cultivate young people, they may no longer rely on the logic of "let them do basic tasks for three years first". Because many basic tasks no longer need to be completed by humans. The new cultivation method may become:

1. AI completes a large number of basic tasks, and young people get exposed to real problems at an earlier stage;

2. Senior employees no longer only assign tasks, but help young people understand why AI gives such answers;

3. Young people not only need to submit answers, but also must explain why they accept or reject AI's answers.

What enterprises need to cultivate is no longer just "skilled employees", but people who can understand the business, raise questions and make judgments as soon as possible.

University education may also need to re-answer a question

This will in turn challenge university education. In the past, the biggest competitive advantage of an excellent university student might be richer knowledge, stronger exam-taking ability, and faster information processing speed. But today, both a fresh graduate and a professional with 20 years of work experience can access almost equally powerful AI.

The gap in knowledge acquisition is narrowing rapidly. At this point, what universities really need to cultivate may increasingly not be "how much knowledge you remember", but:

Can you define a problem?

Can you verify information?

Can you understand the constraints in the real world?

Can you find errors in AI-generated answers?

Can you make reasonable judgments when information is incomplete?

If education still mainly rewards the ability to "give standard answers", while the working world increasingly needs the ability to "judge whether an answer is correct", the gap between universities and enterprises may grow wider and wider. Conclusion: What AI takes away may not be jobs, but the first step on the career ladder.

So looking back at those 1,000 resumes competing for 2 positions, what I am really worried about is not whether these young people can find jobs today. What is more worth thinking about is: through what channels will they gain experience in the future?

The career ladder in the past was very clear: start as an entry-level employee, do basic tasks, gradually become a core member, then grow into an expert and a manager.

AI is quickly dismantling the lowest several steps of this ladder. For people who are already standing in the middle of the stairs, this is good news. They suddenly get a very powerful elevator. A person with 20 years of experience can quickly expand their capability boundaries with the help of AI.

But for young people who have just stood at the entrance of the stairs, the situation is completely different: the first step is disappearing. This may be the real awkwardness of "the most awkward generation in the AI era".

Of course, young people are not without opportunities. Precisely because the value of execution work is declining, they need to complete the transformation that used to take five or even ten years earlier: shift from "what I can do" to "what I can understand, judge and solve".

And enterprises must also take the other half of the responsibility. They cannot reduce entry-level positions while complaining that young people "have no experience". They cannot let AI complete all basic tasks while expecting that there will still be a large number of mature experts in the market ten years later. In the AI era, what really needs to be redesigned may not only be work, but also people's growth paths.